article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
spin models like ising model play a fundamental role in studying phase transitions and critical phenomena in the field of statistical physics and many other disciplines @xcite .
they have also significant implications for understanding social phenomena where co - ordination dynamics is observed , e.g. , in consensus formation and adoption of innovations @xcite .
the spin orientations can represent the choices made by an agent on the basis of information about its local neighborhood .
the two - state majority - vote ( mv2 ) model with noise , proposed by oliveira in 1992 @xcite , is one of the simplest nonequilibrium generalizations of the ising model .
the model presents up - down symmetry and a continuous order - disorder phase transition at a critical value of noise .
studies on regular lattices showed the critical exponents are the same as those of the ising model @xcite , in accordance with the conjecture by grinstein _
the mv2 model has also been extensively studied in complex networks , including random graphs @xcite , small world networks @xcite , scale - free networks @xcite , and some others @xcite . these studies have shown that mv2 models defined on different complex networks belong to different universality classes and the critical exponents depend on the topology of the network topologies . the generalization to a three - state majority - vote model ( mv3 ) on a regular lattice was considered by @xcite , where the authors found the critical exponents for this non - equilibrium mv3 model are in agreement with the ones for the equilibrium three - states potts model @xcite , supporting again the conjecture of @xcite . recently ,
melo _ et al .
_ studied the mv3 model on random graphs and they found that the critical noise is a function of the mean connectivity of the graph @xcite . in @xcite , lima introduced a new zero - state to mv2 model in square lattices and found that this model also falls into the ising universality .
generalized the state variable of the mv model from discrete case to continuous one , and they found that a kosterlitz - thouless - like phase appears in low values of noise @xcite . in the present work ,
we generalize the mv model to arbitrary multiple states where each agent is allowed to be in one of @xmath0 different states with @xmath0 being an integer no less than two . in each step , an agent is randomly chosen and assumes that its state aligns with the state of the majority of its neighborhood with probability @xmath1 and with the minority state with probability @xmath2 . herein , the number of neighboring spins of any agent is fixed that randomly selected from all the other agents at each step .
we mainly focus on the difference in the nature of the order - disorder phase transition between the mv2 model and the mv3 model .
interestingly , we find that in the mv3 model there exists a strong hysteresis behavior as noise intensity goes forward and backward , the main feature of a discontinuous phase transition , in contrast to a continuous phase transition in the mv2 model . a mean - field theory is used to explain the nature of this phase transition .
here , we generalize the original mv model from two states to arbitrary multiple states .
the model is defined as follows .
each agent @xmath3 ( @xmath4 ) can be in any one of @xmath0 states : @xmath5 .
the number of the neighboring agents of agent @xmath3 in any state @xmath6 can be calculated as @xmath7 , where the summation takes over all neighboring nodes of @xmath3 .
@xmath8 is kronecker symbol defined as @xmath9 if @xmath10 and @xmath11 otherwise . with the probability @xmath1 ,
the agent @xmath3 take the same value as the majority spin , i.e. , @xmath12 . with the supplementary probability @xmath2 the agent @xmath3 takes the same value as the minority spin , i.e. , @xmath13 . if there are more than two possible states in the majority spin or in the minority spin , we randomly choose one of them . if @xmath14 , we recover the original mv2 model . in the present work ,
we consider the model defined on random degree - regular networks with annealed randomness , i.e. , the number of neighbors of each agent is fixed but they are randomly chosen from all others at each time step . to characterize the critical behavior of the model , we define the order parameter as @xmath15 where @xmath16 is the imaginary unit .
@xmath17 corresponds to an ordered state , and @xmath18 to a disordered state .
furthermore , we define the susceptibility @xmath19 and binder s fourth - order cumulant @xmath20 as @xmath21\label{eq2}\end{aligned}\ ] ] @xmath22 where @xmath23 denotes time averages taken in the stationary regime .
we have performed extensive monte carlo ( mc ) simulation with various systems of size @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 . at each elementary step , an agent is randomly chosen and tried to update its state according to its neighboring agents as defined before . here , the neighboring agents are selected randomly from all other agents and the number of neighboring agents is fixed at @xmath30 . on each mc step ( mcs ) , each agent is tried to update its state once on average . to obtain steady values of @xmath31 , the first @xmath32 mcs
are discarded and the following @xmath32 mcs are used to calculate time averages of @xmath31 . at the critical region ,
larger runs are performed with @xmath33 mcs to reach the steady state and @xmath34 for computing time averages . on the noise intensity @xmath2 in the mv2 model ( a ) and in the mv3 model ( b ) .
the lines correspond to mean - field results .
[ fig1 ] ] on the noise intensity @xmath2 in the mv2 model ( a ) and in the mv3 model ( b ) .
[ fig2 ] ] in fig.[fig1](a ) and fig.[fig1](b ) , we show @xmath31 as a function of the noise intensity @xmath2 in the mv2 model and in the mv3 model , respectively . for the mv2 model
, one can clearly see that the model displays a continuous order - disorder phase transition as the noise intensity @xmath2 increases .
the maximum in the susceptibility @xmath19 indicates that there exists an effective transition noise @xmath35 for a given @xmath36 , as shown in fig.[fig2](a ) . @xmath35 increases with @xmath36 and such a scaling determines a critical exponent of the continuous phase transition . for the mv3 model , especially for a lager @xmath36 , we find that @xmath31 displays a sharp change close order - disorder phase transition . on the other hand , as shown in fig.[fig2](b )
, the susceptibility @xmath19 also shows a maximum at @xmath35 but the change of @xmath19 near @xmath35 becomes sharper and the values of @xmath19 at @xmath35 becomes larger .
these results may suggest the nature of phase transition in the mv3 model is different from that of the mv2 model . on the noise intensity @xmath2 in the mv2 model ( a ) and in the mv3 model ( b ) .
[ fig3 ] ] to verify our hypothesis , we compare the results of the binder s cumulant @xmath20 in the mv2 model and in the mv3 model . for the mv2 model ,
the phase transition is continuous and the cumulants for different @xmath36 intercept each other at @xmath37 , providing a convenient estimate for the value of the critical noise @xmath38 , as shown in fig.[fig3](a ) . in fig.[fig3](b ) , we plot the curves @xmath39 for different @xmath36 in the mv3 model .
in contrast to the case in mv2 model , @xmath20 exhibits a nonmonotonic dependence on @xmath2 and there exists a minimum near the onset of phase transition . for larger
@xmath36 , @xmath20 falls to negative values near phase transition that is the sign of a discontinuous transition , together with the coexistence of the ordered and disordered phases expected then @xcite .
close the phase transition in the mv3 model .
the parameters are @xmath40 and @xmath41 .
[ fig4 ] ] across a discontinuous phase transition in the mv3 model .
the noise intensity @xmath2 are @xmath42 , @xmath43 , @xmath44 , and @xmath45 from ( a ) to ( d ) , respectively .
the size of the system is @xmath41 .
[ fig5 ] ] to further validate the discontinuous nature of phase transition in the mv3 model , in fig.[fig4 ] we plot a long time series of @xmath31 close the transition noise for @xmath41 .
it is clearly observed that the system flips between the ordered phase and disordered phase due to the finite - size fluctuations .
this implies that at this noise the system is in the coexistence of the ordered and disordered phases , that is a typical of discontinuous phase transition . in fig.[fig5
] , we show the distribution of the values of @xmath31 for several distinct values of @xmath2 across the phase transition in the mv3 model for @xmath41 . for @xmath46 ( fig.[fig5](a ) ) , @xmath31 has a unique maximum around @xmath47 , which corresponds to an ordered phase . for @xmath48 ( fig.[fig5](b ) ) , @xmath31 displays two peaks around @xmath18 and @xmath49 which indicate the coexistence of an ordered phase and a disordered one . upon increasing the value of @xmath2 ,
the peak at @xmath18 is enhanced , while the other peak is depressed ( see fig.[fig5](c ) for @xmath50 ) , implying that the disorder phase becomes more stable .
for even larger values of @xmath2 , there remains a single peak around @xmath18 , which indicates that the system is in a disordered phase ( see fig.[fig5](d ) for @xmath51 ) .
this set of histograms further verify that the mv3 model undergoes a discontinuous order - disorder transition .
note that if the network configuration among agents is quenched other than annealed used in the present work , the mv3 also displays a continuous phase transition as in the mv2 model @xcite .
to analytical understand the difference on the nature of phase transitions between the mv2 model and the mv3 model , we will present a mean - field theory to determine steady values of @xmath31 as a function of @xmath2 . to the end ,
let @xmath52 denote the probability that each node is in the state @xmath6 ( @xmath53 ) .
the time evolution of @xmath52 can be written as @xmath54 where @xmath55 is the transition rate from state @xmath6 to state @xmath56 . according to our model
, the rate @xmath55 can be expressed as the sum of two parts , @xmath57 where the first part is the probability of taking majority - rule @xmath1 by the probability @xmath58 that the state @xmath56 is in the majority state , and the second part is the probability of taking minority - rule @xmath2 by the probability @xmath59 that the state @xmath56 is in the minority state . using the normalization conditions , @xmath60 , eq.([eq4 ] )
can be simplified to @xmath61 in the steady state @xmath62 , we have @xmath63 let @xmath64 denote the number of neighbors in the state @xmath6 among all @xmath65 neighbors , and the probability of a given configuration @xmath66 can be written as multinominal distribution , @xmath67 thus , @xmath68 and @xmath69 can be expressed as @xmath70 and @xmath71 in the present work , the number of neighbors of each agent is kept at @xmath30 . for the mv2 model , @xmath68 and @xmath69 are @xmath72 where @xmath73 and @xmath74 .
one can easily check that @xmath75 ( @xmath76 ) is always a set of solution of eq.([eq4 ] ) for any given @xmath2 .
this trivial solution corresponds to a disordered phase .
the solution loses its stability when the maximal eigenvalue of jacobi matrix is larger than zero , which leads to the analytical value of the critical noise @xmath77 .
the other solutions can be obtained by numerically solving eq.([eq7 ] ) combined with eq.([eq11 ] ) . in fig.[fig6](a ) , we show the steady solutions of @xmath31 as a function of @xmath2 .
the solid and dashed lines indicate the stable and the unstable steady solutions , respectively .
the mv2 model presents a continuous order - disorder phase transition at @xmath78 . for mv3 model , @xmath68 and @xmath69
are @xmath79 where @xmath80 and @xmath81 .
since @xmath82 at @xmath83 ( @xmath84 ) ( corresponding to a disordered phase ) , one can easily check that eq.([eq4 ] ) holds for any given @xmath2 . as mentioned before
, this trivial solution loses its stability when the maximal eigenvalue of jacobi matrix is larger than zero .
it immediately leads to the analytical value of @xmath85 .
the other solutions can be obtained by numerically solving eq.([eq7 ] ) combining with eq.([eq12 ] ) . in fig.[fig6](b ) , we show the steady solutions of @xmath31 as a function of @xmath2 .
the solid line , dashed line and dotted line correspond that the solutions are stable , unstable and saddle , respectively . as @xmath2 varies forward and backward , @xmath31 presents two discontinuous transitions at @xmath86 and at @xmath87 .
this indicates that there exists a hysteresis region at @xmath88 for which the ordered phase and disordered phase are coexisting . in the mv2 model ( a ) and in the mv3 model ( b ) .
the mv2 model displays a continuous order - disorder ( o - d ) phase transition at @xmath78 .
the mv3 model displays a discontinuous o - d phase transition at @xmath86 and a discontinuous d - o phase transition at @xmath89 as @xmath2 varies forward and backward . in the hysteresis ( h ) region between @xmath90 and @xmath91 ,
the mv3 model is in the coexistence of o phase and d phase . the solid line , dashed line and dotted line correspond that the solutions are stable , unstable and saddle , respectively . [ fig6 ] ]
in conclusion , we have generalized the original mv model from two states to arbitrary multiple states .
we have made the detailed comparison on the nature of phase transitions between the mv2 model and the mv3 model defined on annealed networks .
interestingly , we have found that the discontinuous nature of order - disorder phase transition in the mv3 model , qualitatively different from the continuous phase transition in the mv2 model . by a mean - field analysis ,
we have obtained the steady solutions of order parameter @xmath31 as a function of noise @xmath2 .
based on the stability analysis of the solutions , we have obtained the analytical value @xmath91 of noise for the onset of the discontinuous disorder - order phase transition .
another value @xmath90 of noise for the onset of the discontinuous order - disorder phase transition is also numerically estimated . in the hysteresis region between these two transition noises ,
the mv3 model is coexistent of an ordered phase and a disordered one . for a finite size system , the fluctuation can drive the flips between these two phases .
our findings may suggest a novel mechanism of first - order phase transitions in nonequilibrium systems . at last
, we should emphasize that there exists the essential difference in the nature of phase transition between the annealed and quenched mv3 models @xcite .
this work was supported by national science foundation of china ( grants nos .
11205002 , 61473001 , 11475003 , 11405001 ) , the key scientific research fund of anhui provincial education department ( grant no .
kj2016a015 ) and 211 " project of anhui university ( grant no .
j01005106 ) . | in this paper , we generalize the original majority - vote ( mv ) model with noise from two states to arbitrary @xmath0 states , where @xmath0 is an integer no less than two .
the main emphasis is paid to the comparison on the nature of phase transitions between the two - state mv ( mv2 ) model and the three - state mv ( mv3 ) model . by extensive monte carlo simulation and mean - field analysis
, we find that the mv3 model undergoes a discontinuous order - disorder phase transition , in contrast to a continuous phase transition in the mv2 model .
a central feature of such a discontinuous transition is a strong hysteresis behavior as noise intensity goes forward and backward . within the hysteresis region ,
the disordered phase and ordered phase are coexisting . |
in the milestone work by jun kondo , the resistance minimum at nonzero temperature in a metal was explained by the scattering of conduction electrons by localized magnetic impurities @xcite . since that , kondo effect has been considered as a primary mechanism in heavy - fermion systems @xcite . after the description of this effect in the ground state of the kondo lattice model ( klm ) was formulated , the formation of kondo singlet between mobile electron and localized spin is found to perform a key function @xcite .
thanks to the rapid development of laser cooling and trapping techniques , neutral atoms have been considered as an ideal platform for simulating complicated phenomena in condensed matter physics @xcite .
for example , quantum mixtures of alkali atoms are expected to display the characteristics of kondo - correlated state @xcite and multi - orbital effect @xcite . in particular ,
fermionic alkaline - earth atoms ( aeas ) in optical lattices possess unique properties for operating optical atomic clocks with unprecedented precision @xcite and provide novel insights into the physics of strongly correlated transition - metal oxides and heavy - fermion materials as quantum simulators @xcite , where the interorbital spin - exchange interaction between @xmath0s@xmath1 and @xmath2p@xmath1 clock - state atoms plays an essential role @xcite .
active interest has recently been focused on operating neutral atoms as charged particles by engineering artificial gauge potentials @xcite . as a significant breakthrough
, spin hall effect has been observed in a bose - einstein condensate , leading to the realization of a cold - atom spin transistor @xcite , where the internal and external states of atoms are coupled by a two - photon raman process . for trapped atoms in optical lattices ,
tunable gauge potentials have been implemented by periodic driving @xcite and `` zeeman lattice '' @xcite techniques .
furthermore , the harper - hofstadter hamiltonian @xcite has been experimentally realized with ultracold atoms in optical lattices @xcite using laser - assisted tunneling ( lat ) @xcite , leading to a spin - dependent magnetic field and realizing the quantum spin hall ( qsh ) effect @xcite .
moreover , since the interplay between interaction and qsh effect leads to compelling novel physics , both cyclotron motion of repulsive bosons in optical lattices @xcite and qsh insulators with kondo - exchange interaction @xcite in graphene or quantum wells @xcite have been extensively discussed .
motivated by these works , we present a proposal to study the interplay between kondo - exchange interaction and qsh effect with ultracold fermionic aeas trapped in two - dimensional optical lattices using ultracold collision and lat .
the underlying physics is probed by the cyclotron dynamics of a mobile atom and a kondo singlet in a plaquette .
we find that the cyclotron orbit of mobile atom is damped with nonzero kondo coupling .
more interestingly , at strong kondo coupling , we predict a striking and unexpected phenomenon that the spin - zero kondo singlet can behave as a spin - half particle exposed to a perpendicular spin - dependent magnetic flux .
we demonstrate that the spin state of localized background atoms and second - order diagonal tunneling have nontrivial effects on the spin - dependent cyclotron dynamics of the kondo singlet .
( top ) and @xmath3 ( bottom ) @xmath4 atoms . though normal tunneling in the @xmath5 direction is suppressed by a magnetic field gradient @xmath6 for two @xmath4 atom spin states , two raman beams with frequency detuning @xmath7 and momentum transfer @xmath8 restore the resonant tunneling with a spin - dependent complex amplitude @xmath9 .
( b ) schematic picture of effective magnetic flux in a plaquette with four sites @xmath10 and @xmath11 , where the cyclotron orbits of the @xmath4 atom that realize the spin - dependent effective magnetic flux @xmath12 are coupled to a plaquette of localized @xmath13 spins via onsite interorbital spin - exchange interaction ( kondo coupling ) @xmath14 .
@xmath15 denotes the harper hamiltonian . ]
we consider fermionic aeas in @xmath0s@xmath1 ( @xmath4 ) and @xmath2p@xmath1 ( @xmath13 ) clock states independently trapped in two - dimensional optical lattices of the same periodicity @xcite as sketched in fig .
[ fig : model ] , where @xmath16 denotes nuclear spin states @xmath17 @xcite .
the @xmath16 @xmath4 atom ( conduction electron ) receives opposite frequency detuning @xmath18 and momentum transfer @xmath19 from two far - detuned raman beams and tunnels spin - dependently in the optical lattice tilted by a magnetic field gradient @xmath6 along @xmath5 @xcite .
in addition , the mott insulator background of @xmath13 atoms ( localized spins ) interacts with the @xmath4 atom via onsite interorbital spin - exchange interaction @xmath20 , where @xmath21 is the scattering lengths for two atoms in @xmath22 and @xmath23 denotes the wavefunction for the @xmath4 ( @xmath13 ) atom @xcite . by averaging out rapidly oscillating terms in a rotating frame @xcite
, we obtain the time - independent harper - kondo hamiltonian ( see appendix a ) : @xmath24 where @xmath25 and @xmath26 denote the spin operators of the @xmath4 and @xmath13 atoms at ( m , n ) ( @xmath27 and @xmath28 is the vector of pauli matrices ) .
@xmath29 ( @xmath30 ) creates a @xmath4 ( @xmath13 ) atom in the spin state @xmath31 at ( m , n ) .
@xmath32 is the spin - dependent complex tunneling amplitude of the mobile atom in the @xmath5 direction induced by lat , while @xmath33 is the real tunneling amplitude of the mobile atom in the @xmath34 direction .
the first term in eq . ( [ eq : hamiltonian ] ) , i.e. the harper hamiltonian @xmath15 @xcite , describes the @xmath4 atom in a spin - dependent magnetic field arising from spatially varying phase @xmath35 , realizing an effective magnetic flux @xmath36 for @xmath16 @xmath4 atom and leading to spin - orbit coupling and qsh effect @xcite . the last term is the kondo - exchange interaction @xmath37 , which includes the spin - flip process that scatters the @xmath4 atom from @xmath16 to @xmath38 by exchanging the spin with the @xmath38 @xmath13 atom at the same site .
atom ( a ) and the corresponding time evolution of the spin - defect population @xmath39 ( b ) at different kondo - coupling strengths @xmath40 0 ( blue solid ) , @xmath41 ( red dashed ) , @xmath42 ( green dotted ) , and @xmath43 ( purple dash - dotted ) in the period of @xmath44.,scaledwidth=50.0% ] from the cyclotron orbit of noninteracting mobile atom in the period of @xmath44 as a function of kondo coupling @xmath14 . and the transition from the damped cyclotron orbit to the collapsed trajectory happens at @xmath45 , which is marked by the red star.,scaledwidth=50.0% ] we revisit the cyclotron dynamics of the single noninteracting @xmath3 mobile atom in ref .
@xcite when kondo - exchange interaction couples the mobile atom to a plaquette of polarized @xmath46 @xmath13 atoms .
the motion of the mobile atom here can be derived in the hilbert space with quantum number @xmath47 from the state @xmath48 , where @xmath49 denotes the polarized @xmath46 @xmath13 atoms at four sites @xmath10 and @xmath11 associated with the @xmath4 atom vacuum @xcite , and @xmath50 .
@xmath51 is the probability amplitude of @xmath3 @xmath4 atom , and @xmath52 that is relevant to the spin - flip process between @xmath3 @xmath4 atom and @xmath46 @xmath13 atom is the probability amplitude of @xmath3 @xmath13 spin defect in a background of @xmath46 @xmath13 atoms .
we consider the time evolution of the system initially prepared in @xmath53 with @xmath54 and @xmath55 in a period of @xmath56 as ref .
@xcite , and during @xmath44 , the @xmath3 @xmath4 atom without kondo coupling can reproduce the loop trajectory in ref .
@xcite . by solving the schrdinger equation with eq .
( [ eq : hamiltonian ] ) , we obtain the mean position of the @xmath4 atom along @xmath5 and @xmath34 , @xmath57 and @xmath58 , with the onsite population of the @xmath4 atom @xmath59 ( @xmath60 ) and lattice constant @xmath61 .
the trajectories of the mobile atom for different strengths of the kondo coupling @xmath14 are shown in fig .
[ fig : single](a ) . at weak kondo coupling or around quantum critical point ( @xmath62 ) ,
the cyclotron orbit gradually shrinks ; and at strong kondo coupling ( @xmath63 ) , the loop trajectory is totally destroyed , and the atom moves irregularly .
different from cyclotron motion damping induced by repulsive interaction in ref .
@xcite , the damping of the cyclotron orbit of the @xmath4 atom is attributed to the spin - flip process induced by kondo - exchange interaction .
as a consequence of spin flips , the @xmath3 @xmath4 atom evolves into a superposition of @xmath46 and @xmath3 , and since the @xmath46 and @xmath3 @xmath4 atoms manifest opposite chiralities of motion in the spin - dependent magnetic field , the cyclotron orbit shall be evidently damped @xcite . because the spin exchange between @xmath3 @xmath4 atom and @xmath46 @xmath13 atom creates a @xmath3 spin defect in polarized @xmath46 @xmath13 atoms , the time evolution of the spin - defect population @xmath39 in fig .
[ fig : single](b ) monitors the spin - flip event during the motion of the @xmath4 atom . at nontrivial kondo coupling ,
the fractional spin defect population confirms the occurrence of spin flips and the superposition of @xmath46 and @xmath3 @xmath4 atoms ; from @xmath64 to @xmath42 , the spin defect population remarkably increases in the same period as a result of strengthening the spin - flip process ; and the intriguing oscillation at @xmath65 can be considered as a signature of the spin - current vortex in ref .
@xcite , which is related to the formation of composite objects under strong spin flips @xcite .
furthermore , we define the time - averaged interaction - induced deviation from the cyclotron orbit of noninteracting @xmath3 mobile atom in the period of @xmath44 as @xmath66 ^ 2+[\delta y(t)]^2}$ ] , where @xmath67 and @xmath68 .
@xmath69 and @xmath70 are the mean position of the @xmath4 atom along @xmath5 and @xmath34 at time @xmath71 and kondo coupling strength @xmath14 . and a more systematic study of the effect of kondo coupling on the cyclotron orbit is launched by monitoring @xmath72 as a function of @xmath14 in fig .
[ fig : transition ] . at weak kondo coupling or around quantum critical point ,
the deviation @xmath72 increases with the strength of kondo - exchange interaction , indicating that the cyclotron orbit is being further damped , which is consistent with the trajectories derived in the same parameter regime in fig .
[ fig : single](a ) .
the damped cyclotron orbit of the mobile atom begins to collapse when the kondo coupling @xmath14 overwhelms the kinetic energy of the noninteracting mobile atom in the plaquette @xmath73 ( red star ) , where the collapsed trajectory manifests itself by the constant deviation @xmath72 with varying @xmath14 at strong kondo coupling .
@xmath13 atoms . ( b ) the cyclotron trajectories of the kondo singlet in the period of @xmath74 are obtained from @xmath75 ( blue solid ) , @xmath76 ( red dashed ) and @xmath77 ( green dotted ) . if @xmath13 spins become nonparallel , for example , by flipping the @xmath13 spin at b , the cyclotron trajectory disappears ( purple dash - dotted).,scaledwidth=50.0% ] the damping of the cyclotron orbit above shows that the qsh effect of the @xmath4 atom @xcite is not robust to kondo - exchange interaction @xcite . at strong kondo coupling ,
since the @xmath4 and @xmath13 atoms at the same site tend to form a kondo singlet @xcite , we naturally investigate the behavior of a composite object in the synthetic magnetic field .
first we define the creation operators in a background of @xmath46 @xmath13 atoms for the kondo singlet @xmath78 and triplets @xmath79 and @xmath80 , where @xmath81 with @xmath82 is a restriction by the requirement of @xmath47 in the hilbert space .
the physical interpretation of this definition is that the creation of a composite object is accompanied by the annihilation of the @xmath13 atom at the same site .
then , in the singlet - triplet representation , the harper - kondo hamiltonian can be partitioned as @xcite @xmath83 @xmath84 are the projectors onto the subspaces @xmath85 and @xmath86 whose kondo - exchange energy are @xmath87 and @xmath88 , respectively .
considering that the coupling between @xmath89 and @xmath90 is almost energetically forbidden by the large kondo - exchange energy gap @xmath91 , the behavior of the kondo singlet and triplet can be considered to be approximately independent . and
the orientation of the magnetic field can be inferred from @xmath92 of the nearest - neighbor tunneling matrix element @xmath93 in @xmath89 and @xmath90 @xcite . in the high - energy triplet subspace ,
opposite magnetic fields simultaneously work during the tunneling of the triplets : @xmath94 with @xmath95 and @xmath96 with @xmath97 .
consequently , the triplets movement neither realizes effective magnetic flux nor manifests chirality @xcite .
however , in the low - energy singlet subspace , the kondo singlet in a background of @xmath46 @xmath13 atoms experiences a unidirectional magnetic field arising from @xmath98 .
the effective hamiltonian in the low - energy subspace @xmath89 is explicitly given by @xcite ( see appendix b for more details ) @xmath99 \mathcal{p}_t } h_t \mathcal{p}_s \nonumber \\ & = & - \sum_{m , n } \left(\frac{k}{2 } e^{i \phi_{m , n } } s^{\dag}_{m+1,n } s_{m , n } + \frac{j}{2 } s^{\dag}_{m , n+1 } s_{m , n } + h.c.\right ) \nonumber \\ & & - \sum_{m , n } \{j_d \left [ e^{i ( \phi_{m , n } + \phi_{m+1,n } ) } + e^{i ( \phi_{m , n } + \phi_{m , n+1 } ) } \right ] s^{\dag}_{m+1,n+1 } s_{m , n } + h.c.\ } \nonumber \\ & = & h^{(s)}_t + h^{(s)}_d,\end{aligned}\ ] ] where @xmath100 and trivial constants are neglected . the first term in eq . ( [ eq : heff ] ) reconstructs the harper hamiltonian @xmath77 of kondo singlet , and the factor @xmath101 originates from the fact that the effective mass of kondo singlet is twice the mass of the g atom @xcite .
the last term @xmath102 describes the second - order diagonal tunneling of kondo singlet from @xmath103 to @xmath104 or @xmath11 to @xmath105 .
then , @xmath106 reflects the interference between two second - order hopping paths , which reduces to @xmath107 for the paths @xmath108 and @xmath109 . following the procedure above , the time - reversal counterpart of eq .
( [ eq : heff ] ) with @xmath110 can also be obtained in a background of @xmath3 @xmath13 atoms .
this means that the kondo singlet experiences a @xmath13-spin - dependent magnetic field as a spin - half particle , and the opposite chiralities of the kondo singlet s cyclotron orbit shall be locked to polarized @xmath46 and @xmath3 @xmath13 atoms respectively @xcite , thus realizing the qsh effect of the kondo singlet as a result . from the aforementioned demonstrations where the qsh effect of the @xmath4 atom is destroyed at strong kondo coupling and the spin - dependent magnetic field on the kondo singlet
is irrelevant to the spin state of the @xmath4 atom , one can realize that the qsh effect of the kondo singlet revealed in our scheme is nontrivial .
the harper hamiltonian and diagonal tunneling for the kondo singlet in the background of @xmath46 @xmath13 atoms are illustrated in fig .
[ fig : singlet](a ) .
we consider the cyclotron motion of the kondo singlet in the initial state @xmath111 during the period of @xmath74 at @xmath112 . in fig .
[ fig : singlet](b ) , different trajectories of the kondo singlet are calculated by @xmath75 , @xmath76 and @xmath77 .
clearly , the effective hamiltonian @xmath76 at strong kondo coupling resembles the cyclotron dynamics obtained by solving the schrdinger equation with the harper - kondo hamiltonian @xmath75 and accurately describes the cyclotron orbit of the kondo singlet in the plaquette .
furthermore , comparing the cyclotron trajectories obtained by @xmath76 and @xmath77 , we find an essential role of @xmath102 in the movement of kondo singlet .
the singlet travels a longer distance under @xmath76 than that predicted by @xmath77 , because @xmath102 speeds the singlet up around the plaquette .
moreover , the qsh effect of the kondo singlet is not robust to interactions that destroy the parallel alignment of @xmath13 spins . to verify this statement
, we flip one arbitrary spin of the polarized @xmath13 atoms in the initial state , for example , @xmath113 , and find that the cyclotron orbital motion of the kondo singlet disappears .
the result indicates the competition between the qsh effect of the kondo singlet and the interaction - induced nonparallel state of @xmath13 spins @xcite .
now we address practical considerations of observing the cyclotron trajectory of the kondo singlet . here
we only consider ultracold @xmath114yb@xmath115 atoms in the optical lattice .
the raman lasers have to be detuned from the resonant frequency of 399 nm of the @xmath0s@xmath1@xmath116@xmath0p@xmath117 transition . for @xmath0s@xmath1 atoms trapped in a lattice ( @xmath118 nm ) with the depth @xmath119
, we have tunneling rate @xmath120 hz and band gap @xmath121 khz . given land factor @xmath122 and magnetic field gradient @xmath123 mg/@xmath124 m , energy tilting between nearest - neighbor sites is estimated to be @xmath125 hz . therefore , the requirement of @xmath126 can be satisfied @xcite
. the spin - exchange strength @xmath14 on the order of tens of khz @xcite can be tuned by adjusting the lattice depth which changes the overlap of local atomic wavefunctions .
though the @xmath2p@xmath1 atoms can also feel the magnetic field gradient , their wavefunctions are only slightly affected because of the large lattice depth .
the atom distribution can be detected by band - mapping technique in optical superlattices @xcite .
key ingredients in our scheme can be realized by state - of - the - art techniques @xcite .
in summary , we have studied the cyclotron dynamics of the mobile atom and the kondo singlet in the plaquette with interorbital spin - exchange interaction and predicted a novel qsh effect of the spin - zero kondo singlet in a background of polarized spins , demonstrating that the interplay between kondo - exchange interaction and qsh effect constitutes more than competition @xcite for a composite object .
we show that the strong kondo - exchange interaction destroys the loop trajectory of the mobile atom , but instead generates the cyclotron orbital motion for a composite object ( kondo singlet ) . for spin - orbit - coupled heavy fermions in optical square lattices
, our work suggests the competition between the qsh effect of the kondo singlet and the rkky - induced neel state @xcite , which provides opportunities for novel quantum phases and critical points @xcite . as such
, we hope our work may pave a way for experiments on spin - orbit - coupled heavy - fermion systems with aeas in the future .
we thank colin kennedy , hui hu and xia - ji liu for helpful discussions . this work was supported by national foundation of research program ( no .
2011cb921504 ) and nsfc ( no . 11104292 ) .
thanks the support by 100 talents program of cas .
w. thanks the support by open research fund of state key laboratory of precision spectroscopy ( east china normal university ) and nsfc ( no . 11404353 ) .
we extend the scheme of realizing an abelian gauge field with spin - dependent laser - assisted tunneling @xcite to the system considered in the main text as follows : although normal tunneling in the @xmath5 direction is suppressed by a magnetic field gradient @xmath6 for @xmath4 atoms with opposite magnetic moments , a driven raman process with driving frequency @xmath127 and two - photon rabi frequency @xmath128 restores resonant tunneling .
we average out rapidly oscillating terms in a rotating frame and obtain the time - independent harper hamiltonian of the @xmath4 atom : @xmath129 where @xmath130 .
subsequently , we generalize the unitary operator in ref .
@xcite to the spin degree of freedom @xmath131}c^{\dagger}_{m , n,\alpha}c_{m , n,\alpha},\ ] ] and derive the expression of kondo - exchange interaction in the above rotating frame at resonant tunneling @xmath132 \nonumber \\ & = & v_{ex } \sum_{m , n } s^{c , z}_{m , n } s^{f , z}_{m , n } + \frac{1}{2 } j_0 \left ( \frac{2\omega}{\delta}\sin\phi_{m , n } \right ) ( s^{c+}_{m , n}s^{f-}_{m , n}+s^{c-}_{m , n}s^{f+}_{m , n}),\end{aligned}\ ] ] where @xmath133 are the bessel functions of the first kind
. the anisotropy of the term @xmath134 can often be relaxed .
when @xmath135 , because @xmath136 , the anisotropy for an arbitrary phase @xmath95 can be reasonably ignored .
thus , in future experiments , by choosing a proper ratio between two - photon rabi frequency and magnetic field gradient , the isotropic kondo - exchange interaction can always be reached in the driven optical lattice : @xmath137
in the main text , we obtain the effective hamiltonian in the strong kondo - coupling regime by the projection @xmath138
\mathcal{p}_t } h_t \mathcal{p}_s . \nonumber\ ] ] here , we divide @xmath76 into two parts @xmath139 and @xmath140 , and the matrix elements of @xmath141 are explicitly given in the low - energy singlet subspace @xmath142 where @xmath143 represents plaquette sites @xmath10 and @xmath11 .
@xmath144 includes only the nearest - neighbor tunneling term @xmath145 which gives @xmath77 in the main text .
@xmath146 includes the onsite interaction , as well as the nearest - neighbor and diagonal tunneling terms .
first , the onsite interaction term only contributes the trivial constant @xmath147 second , the nearest - neighbor tunneling term can be safely neglected because @xmath148 third and last , we obtain the second - order diagonal tunneling as @xmath149 where @xmath150 originates from the interference between possible paths of diagonal tunneling from @xmath103 to @xmath104 or @xmath11 to @xmath105 .
for example , a kondo singlet initially at site @xmath103 can pass through @xmath105 or @xmath11 to arrive at diagonal site @xmath104 . in the tunneling path @xmath108 , the singlet carries a zero phase because @xmath151 and @xmath152 .
however , in the other path @xmath109 , the accumulated phase changes to @xmath153 . as a result
, the interference between two different tunneling paths contributes @xmath154 .
finally , @xmath146 gives @xmath102 in the main text .
d. l. cox and a. zawadowski , adv .
phys . * 47 * , 599 ( 1998 ) . i. bloch , j. dalibard and s. nascimbne , nat .
* 8 * , 267 ( 2012 ) .
j. bauer , c. salomon , and e. demler , phys .
lett . * 111 * , 215304 ( 2013 ) .
y. nishida , phys .
* 111 * , 135301 ( 2013 ) .
a. derevianko and h. katori , rev .
phys . * 83 * , 331 ( 2011 ) .
b. j. bloom , t. l. nicholson , j. r.williams , s. l. campbell , m. bishof , x. zhang , w. zhang , s. l. bromley , and j. ye , nature * 506 * , 71 ( 2014 ) .
a. v. gorshkov , m. hermele , v. gurarie , c. xu , p. s. julienne , j. ye , p. zoller , e. demler , m. d. lukin , and a. m. rey , nat . phys .
* 6 * , 289 ( 2010 ) .
m. foss - feig , m. hermele , and a. m. rey , phys .
a * 81 * , 051603(r ) ( 2010 ) .
k. r. a. hazzard , v. gurarie , m. hermele , and a. m. rey , phys .
a * 85 * , 041604(r ) ( 2012 ) .
s. sugawa , k. inaba , s. taie , r. yamazaki , m. yamashita , and y. takahashi , nat . phys .
* 7 * , 642 ( 2011 ) .
s. taie , r. yamazaki , s. sugawa and y. takahashi , nat . phys .
* 8 * , 825 ( 2012 ) .
x. zhang , m. bishof , s. l. bromley , c. v. kraus , m. s. safronova , p. zoller , a. m. rey , and j. ye , science * 345 * , 1467 ( 2014 ) .
f. scazza , c. hofrichter , m. hfer , p. c. de groot , i. bloch , and s. flling , nat . phys . * 10 * , 779 ( 2014 ) .
g. cappellini , m. mancini , g. pagano , p. lombardi , l. livi , m. siciliani de cumis , p. cancio , m. pizzocaro , d. calonico , f. levi , c. sias , j. catani , m. inguscio , l. fallani , phys .
lett . * 113 * , 120402 ( 2014 ) .
j. dalibard , f. gerbier , g. juzelinas , and p. hberg , rev .
phys . * 83 * , 1523 ( 2011 ) .
lin , r. l. compton , a. r. perry , w.d .
phillips , j.v .
porto , and i. b. spielman , phys .
lett . * 102 * , 130401 ( 2009 ) .
lin , r. l. compton , k. jimnez - garca , j. v. porto , and i. b. spielman , nature * 462 * , 628 ( 2009 ) .
p. wang , z .- q .
yu , z. fu , j. miao , l. huang , s. chai , h. zhai , and j. zhang , phys .
* 109 * , 095301 ( 2012 ) .
l. w. cheuk , a. t. sommer , z. hadzibabic , t. yefsah , w. s. bakr , and m. w. zwierlein , phys .
lett . * 109 * , 095302 ( 2012 ) .
m. c. beeler , r. a. williams , k. jimnez - garca , l. j. leblanc , a. r. perry , and i. b. spielman , nature * 498 * , 201 ( 2013 ) .
j. struck , c. olschlger , m. weinberg , p. hauke , j. simonet , a. eckardt , m. lewenstein , k. sengstock , and p. windpassinger , phys .
lett . * 108 * , 225304 ( 2012 ) .
k. jimnez - garca , l. j. leblanc , r. a. williams , m. c. beeler , a. r. perry , and i. b. spielman , phys .
lett . * 108 * , 225303 ( 2012 ) .
p. harper , proc .
london sect .
a * 68 * , 874 ( 1955 ) .
m. aidelsburger , m. atala , m. lohse , j. t. barreiro , b. paredes , and i. bloch , phys .
lett . * 111 * , 185301 ( 2013 ) .
h. miyake , g. a. siviloglou , c. j. kennedy , w. c. burton , and w. ketterle , phys .
* 111 * , 185302 ( 2013 ) .
a. r. kolovsky , europhys .
93 * , 20003 ( 2011 ) ; c. e. creffield and f. sols , europhys .
* 101 * , 40001 ( 2013 ) . c. j. kennedy , g. a. siviloglou , h. miyake , w. c. burton , and w. ketterle , phys .
* 111 * , 225301 ( 2013 ) .
m. aidelsburger , m. lohse , c. schweizer , m. atala , j. t. barreiro , s. nascimbne , n. r. cooper , i. bloch , and n. goldman , arxiv : 1407.4205v1 ( 2014 ) .
xiaopeng li and s. das sarma , phys . rev .
b * 89 * , 224302 ( 2014 ) . c. wu , b. a. bernevig , and s .- c . zhang , phys .
lett . * 96 * , 106401 ( 2006 ) .
j. maciejko , c. liu , y. oreg , x .-
qi , c. wu , and s .- c .
zhang , phys .
* 102 * , 256803 ( 2009 ) .
j. maciejko , t. l. hughes , and s .- c .
zhang , annu .
. condens .
matter phys .
* 2 * , 31 ( 2011 ) .
qi , and s .- c .
zhang , rev .
phys . * 83 * , 1057 ( 2011 ) .
f. goth , d. j. luitz , and f. f. assaad , phys .
b * 88 * , 075110 ( 2013 ) .
j. maciejko , phys .
b * 85 * , 245108 ( 2012 ) .
feng , j. dai , c .- h .
chung , and q. si , phys .
. lett . * 111 * , 016402 ( 2013 ) .
a. j. daley , m. m. boyd , j. ye , and p. zoller , phys .
lett . * 101 * , 170504 ( 2008 ) ; a. j. daley , j. ye , and p. zoller , eur .
j. d * 65 * , 207 ( 2011 ) .
s. methfessel and d. c. mattis , handbuch der physik , vol .
18 ( springer , berlin , 1968 ) ; | we propose a scheme to investigate the interplay between kondo - exchange interaction and quantum spin hall effect with ultracold fermionic alkaline - earth atoms trapped in two - dimensional optical lattices using ultracold collision and laser - assisted tunneling . in the strong kondo - coupling regime , though the loop trajectory of the mobile atom disappears , collective dynamics of an atom pair in two clock states can exhibit an unexpected spin - dependent cyclotron orbit in a plaquette , realizing the quantum spin hall effect of the kondo singlet .
we demonstrate that the collective cyclotron dynamics of the spin - zero kondo singlet is governed by an effective harper - hofstadter model in addition to second - order diagonal tunneling . |
laws of electrical circuits are drastically modified when mesoscopic systems are incorporated .
a coherent conductor embedded into a circuit sees the imposed voltage by the generator reduced by a fluctuating voltage associated with the impedance of the surrounding electromagnetic environment .
this gives rise to a current reduction with a pronounced non - linearity , called zero - bias anomaly ( zba ) : transfer of electrons becomes inelastic as they exchange photons with the electromagnetic environment .
this phenomena , called the dynamical coulomb blockade ( dcb ) , has attracted a tremendous interest both theoretically and experimentally.@xcite nevertheless , most of the works have been initially developed by focusing on the limit of a weakly transmitting conductor , i.e. , the tunneling regime.@xcite more recently , interest in the regime of few well - transmitting channels has emerged . on one hand , reducing the number of channels makes more apparent the effect of the circuit , as many channels could play as well the role of an out - of - equilibrium environment .
on the other hand , highly transmitting conductors raise two interesting questions .
first , whether the charge fluctuations wash out the dcb , and second , whether the reduction in the current is related to shot noise in the absence of the environment .
indeed , this intuitive and attractive relation had been proposed through a perturbative computation with respect to a very weak impedance , and has been checked experimentally.@xcite nevertheless , its validity domain is restricted to high enough energies , as logarithmic divergences arise : one needs to go beyond the weak feedback action .
this was particularly the case for an ohmic environment , i.e. , having an impedance @xmath4 at frequencies @xmath5 ( where @xmath6 is the resistance and @xmath7 the capacitance ) : this situation has not only relevance to realistic experiments but also a fundamental interest , being related to the investigation of electronic interactions .
the logarithmic divergences have been resumed using a renormalization group ( rg ) scheme by kindermann and nazarov,@xcite dealing again with @xmath8 , where @xmath0 is the quantum of resistance .
it has been nevertheless possible , in a simultaneous and independent work , to deal for the first time with an arbitrary value of @xmath6 : one of the authors , with saleur , @xcite has shown that the problem of a short - coherent conductor in series with a resistance @xmath6 is equivalent to the impurity problem in a tomonaga - luttinger liquid@xcite ( tll ) with an interaction parameter : @xmath9 when specified to the limit of small @xmath6 , this mapping had led to recover the same results as those by kindermann and nazarov .
these findings answer the two questions addressed above .
first , the dcb still persists at good transmission , showing up below an energy scale @xmath10 ( depending in a non - universal way on @xmath6 and on transmission ) : it corresponds to the crossover voltage between the so - called weak backscattering ( wbs ) regime at high energy and strong backscattering ( sbs ) regime at low energy.@xcite secondly , the reduction in the current is related to the noise in the presence of the environment , and not to the noise of the isolated conductor as stated before ( see eq .
( [ crucial ] ) ) .
recent pioneering experiments by pierre s group@xcite where the strong feedback of an arbitrary impedance on a one - channel edge state has been investigated for the first time has shown satisfactory agreement with the theoretical predictions of refs . and .
interestingly , even though one has to take strictly the limit of one channel in ref .
, the equivalence to a tll thus established seems to extend to many channels when they are treated in a mean - field framework , provided one renormalizes the parameter @xmath11 by including the resistance of the channels , and scales the voltage appropriately , as one can infer from a more recent study.@xcite by mapping a one - channel conductor in series with an ohmic environment to a tll , the parameter @xmath11 of which can be controlled by tuning @xmath6 ( see eq .
( [ key ] ) ) , one gets as well a promising alternative to test the theoretical predictions for the impurity problem in a tll . indeed , satisfactory experimental evidence has been lacking , even though partial success has been claimed to be obtained . in particular , as noticed in ref .
, the case of an environmental resistance equal to the quantum of resistance , @xmath12 , corresponds to a tll with an interaction parameter @xmath2 . in that case
, the impurity problem in a tll can be solved exactly in a more transparent way compared to other values of @xmath11 ( see ref .
for instance ) .
such an apparent simplicity is due to the introduction of new chiral fermions which arise from a mathematical construction without any physical entity , and incorporate non - trivial strong correlation effects .
the non - perturbative investigation has a fundamental interest : it explores crucial issues not reached through perturbative approaches .
nevertheless , the situation with @xmath2 has never been achieved experimentally .
indeed , the ideal candidate to investigate the tll s behavior is usually provided by edge states with a constriction in the fractional quantum hall effect ( fqhe ) . however , as @xmath11 plays the role of the fractional filling factor @xmath13 , and as the description of the edges in terms of one - channel chiral tll model is valid only for simple values : @xmath14 with @xmath15 an integer , it has not been possible to achieve a chiral tll with @xmath2 .
other potential candidates are provided by quantum wires and carbon nanotubes .
however three main difficulties arise in those systems :
( i ) achieving only one backscattering center ; ( ii ) tuning the interaction parameter @xmath11 ; and ( iii ) taking into account the connection to reservoirs and finite size effects , where only a perturbative treatment of the impurity has been possible.@xcite thus , a coherent one - channel conductor connected to a quantum of resistance @xmath12 , as achieved recently,@xcite offers an unique opportunity to explore the properties of a tll with an impurity at @xmath2 .
the specific value , @xmath2 , is exciting as it allows us to obtain handy analytic expressions for dc current and the zero - frequency noise . beyond the stationary regime ,
time - dependent transport probes even more the dynamics and tests in a precise way the underlying model .
this is the case of finite - frequency ( ff ) noise . in view of the mapping in ref .
, one can use previous results obtained in the fqhe . on the one hand , as far as the ff symmetrized noise is concerned , chamon _
et al._@xcite have computed it perturbatively with respect to the impurity strength , and non - perturbatively at the specific value @xmath2 . on the other hand , within the framework of the exact solution of ref . ,
lesage and saleur@xcite have obtained only its behavior for low frequencies or close to the `` josephson '' singularity @xmath16 , where @xmath17 denotes the voltage and @xmath18.@xcite notice that both approaches disagree , apart from the particular value @xmath2 .
nevertheless , it has been possible to measure experimentally the ff non - symmetrized noise@xcite , which is more interesting to explore .
it can be inferred from ref .
dealing with edge states at simple filling factors , both in the wbs and sbs regimes : the same results apply to a coherent conductor with a good bare transmission @xmath19 connected to an arbitrary resistance at high or low enough energies , as well as for a low transmission.@xcite our aim is to go beyond this perturbative computation , offering a full description extending over all energy ranges below @xmath20 , where @xmath21 is the frequency cutoff associated to the fermionic degrees of freedom in the conductor .
this is precisely possible at @xmath12 .
thus , our study gives the first non - perturbative results for the ff non - symmetrized noise , without assuming neither weak resistance , nor high or low enough energies associated to the wbs nor sbs regimes .
in addition , this offers a benchmark for other values of the resistance .
one of the key steps within the refermionization procedure is that the chiral independent fermions have now an energy - dependent transmission amplitude @xmath22 ( see eq .
( [ tomega ] ) ) which encodes the non - trivial many - body correlations . indeed , the associated transmission coefficient @xmath23 has nothing to do with the effective transmission obtained for @xmath24 by the rg approach,@xcite and can not be obtained adiabatically from that in the absence of interactions , @xmath19 .
it has been widely accepted that transport properties of the system can be obtained within the scattering approach using @xmath22 .
this was shown to be valid both for the dc current , as well as for the full counting statistics ( fcs).@xcite surprisingly , our study shows that the scattering approach fails when one deals with time - dependent transport : the ff noise can still be expressed in terms of @xmath22 , nevertheless it obeys a different relation .
the same feature occurs for another quantity : the out - of - equilibrium ff dissipative conductance @xmath25 $ ] which depends on the applied dc voltage and the frequency of the superimposed modulation , in addition to its implicit dependence on temperature .
this last quantity has started to be studied only recently in few correlated systems , where its interest has been shown : quantum wires with an impurity@xcite and kondo problem.@xcite this has become possible owing to a crucial result : @xmath26 $ ] is given by an out - of - equilibrium kubo type formula , i.e. , by the asymmetry between the emission and absorption ff noise.@xcite in this present and first non - perturbative investigation , we show that @xmath26 $ ] does not fit with its expression obtained within the scattering approach .
even more , it obeys exactly a surprising relation , eq .
( [ eq_g_i ] ) , which determines it fully from the dc current , in a way similar to that obtained in tunnel junctions within the tien - gordon theory , but with a renormalized charge here.@xcite recently , such a relation has been generalized to full extent , including arbitrary interactions : it has been shown to be universal to lowest order with respect to a local or spatially extended tunneling at arbitrary dimension , as well as with respect to local or spatially extended backscattering if one - dimensional systems are considered.@xcite notice that our results will depend on parameters such as the frequency cutoff @xmath27 , an effective transmission we call @xmath28 , and the voltage scale @xmath29 , which is related in a non - universal way to @xmath27 and @xmath28 .
nevertheless , we will show that the scaling laws , known to be obeyed by the dc current , can be extended to the ff conductance and ff noise .
more precisely , all these quantities depend only on @xmath30 , @xmath31 and @xmath32 , where @xmath33 denotes the temperature .
.synthesis of the abbreviations .
[ cols= " < , < " , ] the important effective parameter is indeed the scaling voltage @xmath29,@xcite which appears in the bethe - ansatz solution as non - universal.@xcite it can be roughly related to the backscattering amplitude through @xmath34 .
indeed @xmath29 characterizes the crossover between the wbs and sbs regimes . at energies high enough compared to @xmath10 , one can use the perturbative rg analysis with respect to a weak impurity by kane and fisher.@xcite in the opposite limit ( i.e. , at energies @xmath35 ) the wire is cut into two pieces , with weak tunneling between them .
thus , even though one starts from a weak impurity , lowering the energy drives the system from the wbs behavior , where the conductance is slightly reduced , into the sbs regime where the conductance is suppressed .
this means that the dcb still takes place even when the effective transmission @xmath28 is close to one,@xcite but below an effective charging energy @xmath10 .
in particular , at energies much smaller than @xmath10 , thus in the sbs regime , it has been shown that one recovers the @xmath36 theory which is rather obtained starting from a very weak transmission.@xcite even more , it is possible to obtain non - perturbative results not only with respect to @xmath6 , but also to @xmath37 , and to describe the whole regime of energies below @xmath38 .
such an achievement was made possible by exploiting the bethe - ansatz exact solutions of fendley _
et al._@xcite for the impurity problem in a tll . within the bethe - ansatz solution ,
the fcs can be computed exactly as well .
in particular , using these results for the current and noise , it was possible to derive a crucial and exact relation between the derivative of the differential conductance @xmath39 and the differential zf noise @xmath40 at zero temperature : @xmath41 let us remark that in the limit of small @xmath8 , eq .
( [ crucial ] ) fits perfectly with the rg equation obtained by kindermann and nazarov.@xcite within the bethe - ansatz solution , the fcs can be computed exactly as well . using its derivation at zero temperature,@xcite the mapping permits to extend eq .
( [ crucial ] ) to higher cumulants .
it has also motivated partly the recent investigations of the fcs at finite temperature.@xcite more recently , in an interesting work , golubev _ et al._@xcite have studied a mesoscopic interacting multi - channel conductor connected to an ohmic environment , using a method based on keldysh action .
they have confirmed the results presented above ( of ref . ) in the perturbative regime they restrict to ( with respect to @xmath37 ) , which corresponds to high energies in the wbs domain.@xcite even more , when treating many channels in a mean - field approach , the agreement holds too , provided the tll s parameter in eq .
( [ key ] ) incorporates the resistance of the conductor .
this shows that the mapping can be extended to many channels , as well as its consequences discussed above ( at least at high enough energy ) , and motivates further the computation done in this work . in the following
, we will consider the situation where the resistance is equal to the quantum of resistance : @xmath12 . in that case , the tll parameter is @xmath2 ( see eq .
( [ key ] ) ) , and eq . ( [ hamilton ] ) is exactly solvable through a refermionization procedure,@xcite thus one can perform a non - perturbative analysis of transport properties .
the refermionization introduces new independent chiral fermions with a frequency dependent transmission amplitude : @xmath42 thus a transmission coefficient : @xmath43 where @xmath44 is the energy crossover between the wbs and sbs regimes .
the profile of @xmath45 is shown in the lower panel of fig .
[ figure1 ] . obviously , @xmath45 is identical to @xmath46 ( perfect effective transmission ) when @xmath47 whatever the frequency is . as we will show later on
, @xmath48 yields the non - linear differential conductance @xmath49 at zero temperature . in accordance with the features concerning the crossover from wbs to sbs
, one sees that as soon as @xmath28 deviates from one , @xmath45 decreases quickly to zero at low frequency compared to @xmath50 . from eq .
( [ eq_t ] ) , notice that @xmath51 coincides with the effective transmission @xmath28 .
in this section , we present the formal results for the dc current , the non - symmetrized noise , and the differential conductance .
we first calculate the dc current , defined as the average of the time derivative of the charge which is transferred through the conductor : @xmath52 , where @xmath53 is the current operator . since we consider a dc applied voltage , @xmath54 is time - independent .
the details of the calculation are presented in appendix [ app_current ] .
we obtain:@xcite @xmath55~,\end{aligned}\ ] ] which corresponds to the landauer formulation of the current , where the fermi - dirac distribution function is given by @xmath56^{-1}$ ] .
the density of states multiplied by the velocity is a constant , equal to @xmath57 , because the energy spectrum of the new fermions is linear too .
( [ eq_current ] ) describes the behavior of the dc current over all voltage and temperature ranges , starting from the wbs regime down to the sbs regime .
we have to recall that the model describes a strongly correlated system where interactions can not be treated by any mean - field approach , and that the transmission amplitude incorporates their non - trivial effects .
notice that in eq .
( [ eq_current ] ) we have made the choice of extending the limits of integration to plus and minus infinity .
strictly speaking , these limits should be @xmath58 and @xmath27 , however we have checked that the correction terms are negligible . in the following , a similar choice will be made in all integrals over frequencies .
next , we calculate the ff non - symmetrized noise defined as the fourier transform of the current fluctuations : @xmath59 where @xmath60 .
we obtain the following result ( see appendix [ app_noise ] for details ) : @xmath61[1-f(\hbar\omega'\pm ev/2)]f(\hbar\omega'+\hbar\omega\pm ev/2)\nonumber\\ & & + \big[\mathcal{t}(\omega')-\mathcal{t}(\omega')\mathcal{t}(\omega+\omega')-|t(\omega')-t(\omega+\omega')|^2/4\big][1-f(\hbar\omega'\pm ev/2)]f(\hbar\omega'+\hbar\omega\mp ev/2)\bigg)~.\end{aligned}\ ] ] a crucial and surprising observation is that this expression differs from that obtained within the scattering approach for a one - channel conductor with an energy - dependent transmission@xcite ( see appendix [ app_comp ] for more details on that comparison ) , even when applied to chiral fermions.@xcite it is interesting to cast eq .
( [ eq_noise ] ) under the alternative form : @xmath62 \nonumber\\ & & + \frac{e^2}{4\pi}\sum_\pm [ n(\hbar\omega\pm ev)-n(\hbar\omega ) ] \nonumber\\ & & \times\int_{-\infty}^{\infty}d\omega ' \left[\mathcal{t}(\omega')\mathcal{t}(\omega+\omega')+\frac{|t(\omega')-t(\omega+\omega')|^2}{4}\right]\nonumber\\ & & \times[f(\hbar\omega+\hbar\omega'\pm ev/2)-f(\hbar\omega'\mp ev/2)]~,\end{aligned}\ ] ] where @xmath63^{-1}$ ] is the bose - einstein distribution function .
notice that in the trivial limit @xmath64 ( i.e. , perfect transmission ) , the voltage drop across the mesoscopic conductor is @xmath65 and the ff noise reduces to @xmath66 whatever the temperature is , where @xmath67 is the quantum of conductance .
an interesting point is that , in the r.h.s . of eq .
( [ eq_noise_bis ] ) , both distribution functions @xmath68 for fermions and @xmath69 for bosons ( of the electromagnetic environment , thus electron - hole excitations ) are involved , which explains the fact that the effective voltage is not the same in their arguments : one has @xmath70 for the function @xmath68 , and @xmath71 for the function @xmath69 . from eq .
( [ eq_noise ] ) , we deduce immediately the zf noise : @xmath72f(\hbar\omega'\pm ev/2)\nonumber\\ & & + \mathcal{t}(\omega')\left(1-\mathcal{t}(\omega')\right)[1-f(\hbar\omega'\pm ev/2)]f(\hbar\omega'\mp ev/2)\big)~.\nonumber\\\end{aligned}\ ] ] contrary to what occurs for the ff noise , one recovers , in the zf limit , an expression in terms of the transmission coefficient @xmath73 similar to that within the scattering theory .
let us now calculate the ff conductance which is related to the non - symmetrized noise through the exact relation:@xcite @xmath74=\frac{s(v,-\omega)-s(v,\omega)}{2\hbar\omega}~.\end{aligned}\ ] ] reporting eq .
( [ eq_noise ] ) in eq .
( [ def_g ] ) , all the terms which contain a product of transmission coefficients vanish , and we find simply : @xmath75=\frac{e^2}{4h\omega}\sum_\pm\int_{-\infty}^{\infty}d\omega ' \mathcal{t}(\omega')\nonumber\\ & & \times\big[f(\hbar\omega'\pm ev/2)-f(\hbar\omega+\hbar\omega'\pm ev/2)\big]~,\end{aligned}\ ] ] which obeys unexpectedly the exact relation:@xcite @xmath76&=&\frac{e}{4\hbar\omega}\left[i\left(v+2\hbar\omega / e\right)-i\left(v-2\hbar\omega / e\right)\right]~.\nonumber\\\end{aligned}\ ] ] this is a central result of our paper : even though our computation is non - perturbative , one recovers a similar relation to that obtained within the perturbative tien - gordon theory , with a renormalized charge @xmath77.@xcite indeed , such a relation has been recently shown in ref . to be generally valid for all strongly correlated systems at arbitrary dimensions , with possible coupling to an arbitrary electromagnetic environment , provided
one requires the tunneling regime . in one - dimensional systems ,
the same work has shown the universality of the relation for one or many weak impurities in the wbs regime aside from its validity in the sbs regime or for tunneling barriers . here , surprisingly , for a tll with an impurity and @xmath2 , or a conductor coupled to a quantum resistance , eq .
( [ eq_g_i ] ) is valid for all voltage , temperature and frequency ranges ( below @xmath78 ) with a renormalized charge @xmath77 .
it is interesting as well to specify to frequencies much larger than the applied voltage @xmath17 , where the ff conductance becomes voltage - independent , thus reaches a linear regime . in that case , eq .
( [ eq_g_i ] ) can be approximated simply , as in ref .
, by : @xmath79\approx \frac{ei(2\hbar\omega / e)}{\hbar\omega}~.\end{aligned}\ ] ] in the two following sections , starting from the expressions derived here , we first recall some known results of the stationary regime ( sec .
iv ) and next , we discuss in details the news results of the time - dependent regime ( sec .
in this section , we explicit in more details the dc transport in all the temperature and voltage ranges , starting from the low temperature regime ( @xmath80 ) and ending with the high temperature regime ( @xmath81 ) .
_ low temperature behavior_. in the limit of strictly zero temperature , the integral over frequency in eq .
( [ eq_current ] ) can be performed analytically and the dc current reads as:@xcite @xmath82~.\end{aligned}\ ] ] the dc current is thus given by a product of @xmath29 multiplied by a function of @xmath30 .
this result has been used successfully to describe recent experimental data.@xcite for perfect effective transmission , @xmath47 ( i.e. , @xmath83 ) , we recover ohm s law : @xmath84 , whereas at @xmath85 , the dc current is reduced : dcb persists even though one starts from a good effective transmission . even more , one can check , as shown in ref . , that in the sbs regime , at @xmath86 , one recovers a power law behavior controlled by the exponent @xmath87 as in the @xmath36 theory , and obtained rather for a weakly transmitting conductor.@xcite for that , one expands eq .
( [ eq_current_t_0 ] ) with respect to @xmath30 : @xmath88 in the opposite limit of the wbs regime , i.e. at @xmath89 , the reduction to the perfect current , called the backscattering current : @xmath90 can be expanded with respect to @xmath91 : @xmath92 notice that the first term on the r.h.s .
corresponds to that obtained within the perturbative computation,@xcite @xmath93 , which becomes voltage independent at @xmath2 .
, as a function of @xmath94 , for different values of @xmath28 , at @xmath95 .
since we have @xmath96 , the corresponding values of @xmath29 are : @xmath97 ( red solid line ) , @xmath98 ( green dashed line ) , @xmath99 ( blue short dashed line ) , and @xmath100 ( black dotted line ) .
right panel : all the curves of the left graph scale to a single one when one considers the variation with @xmath30 .
we take @xmath101 .
[ figure2],title="fig:",width=158 ] , as a function of @xmath94 , for different values of @xmath28 , at @xmath95 .
since we have @xmath96 , the corresponding values of @xmath29 are : @xmath97 ( red solid line ) , @xmath98 ( green dashed line ) , @xmath99 ( blue short dashed line ) , and @xmath100 ( black dotted line ) .
right panel : all the curves of the left graph scale to a single one when one considers the variation with @xmath30 .
we take @xmath101 .
[ figure2],title="fig:",width=158 ] the differential conductance can be obtained either by letting @xmath102 in eq .
( [ eq_g ] ) , or by differentiating the dc current of eq .
( [ eq_current_t_0 ] ) . at zero temperature
, it reads as : @xmath103~,\end{aligned}\ ] ] and obeys the relation : @xmath104 where the behavior of the transmission coefficient @xmath73 is shown on the lower panel of fig .
[ figure1 ] .
notice that , at @xmath47 , the differential conductance becomes constant ( linear ) : it is equal to @xmath105 as the ballistic conductor is in series with a resistance @xmath12 . at @xmath106
, @xmath49 can formally reach @xmath105 for @xmath107 , nevertheless it stays below since @xmath3 is limited by the cutoff @xmath78 .
the left panel of fig .
[ figure2 ] gives the differential conductance at low temperature , which shows a zba that is more pronounced when @xmath28 is reduced .
a surprising result is its large sensitivity with respect to small variations of @xmath28 in the vicinity of @xmath46 ( see the red and green curves in fig .
[ figure2 ] for example ) : it is due both to the rapid variations of @xmath108 at frequencies low compared to @xmath27 ( when @xmath28 is close to @xmath46 , see the lower panel of fig . [ figure1 ] ) which are precisely those that give the dominant contribution to the integral over @xmath109 in eq .
( [ eq_g]).@xcite we can see that the crossover @xmath29 varies rapidly in the vicinity of @xmath28 close to one , thus the sbs regime is reached much more rapidly when @xmath28 decreases .
it is important to recall that all the curves scale to the same one when one scales the voltage by @xmath29 , as shown in the right panel of fig .
[ figure2 ] . , for @xmath110 , at a low temperature :
the arrows indicate the associated values of @xmath111 . with a frequency cutoff of about @xmath46 thz
, the value @xmath112 @xmath113 on the axis corresponds to a current of about @xmath114 na.,width=226 ] at zero - temperature , the zf noise of eq .
( [ eq_zero_noise ] ) becomes : @xmath115 as for the dc current , the zf noise is given by a product of @xmath29 multiplied by a function of @xmath30 .
one can check that the zf noise obeys eq .
( [ crucial ] ) , with @xmath12 , which confirms that the zba is exactly linked to the zf shot - noise in the presence of the electromagnetic environment .
figure [ figure3 ] shows the dc current and the zf noise as functions of @xmath28 , at low temperatures compared to voltage , @xmath80 : the current shows an increasing behavior , whereas the zf noise is non monotonous .
this is due to the fact that the current is expressed as the integral of the transmission coefficient , see eq .
( [ eq_current ] ) , and that the integral expression of the zf noise of eq .
( [ eq_zero_noise ] ) contains a term proportional to @xmath116 . for @xmath117 ( i.e. , @xmath118 using the parameters of fig .
[ figure3 ] ) , the current and zf noise curves converge to the same value .
thus , the fano factor given by the ratio @xmath119 tends to @xmath120 , the value obtained in the poissonian regime with independent transfer events of charge @xmath120 trough the mesoscopic conductor . for @xmath121 ,
the transfer events are no more independent but the noise is still close to zero because of the @xmath122 factor .
in addition , the fano factor defined this time by the ratio @xmath123 , where @xmath124 is the backscattering current given by eq .
( [ back ] ) is equal to @xmath125 . in the fqhe at @xmath14 , the fano factor would be given by fractional charge @xmath18 .
in the dcb context , this renormalization is related to the dc conductance @xmath105 without backscattering .
, as a function of temperature @xmath33 , in units of @xmath126 , for different values of @xmath28 , at @xmath127 .
right panel : all the curves of the left graphic scale to a single one when one considers the variation with respect to @xmath31 .
[ figure4],title="fig:",width=158 ] , as a function of temperature @xmath33 , in units of @xmath126 , for different values of @xmath28 , at @xmath127 .
right panel : all the curves of the left graphic scale to a single one when one considers the variation with respect to @xmath31 . [ figure4],title="fig:",width=158 ] _ intermediate temperature behavior_. the left panel of fig .
[ figure4 ] shows the differential conductance at zero voltage as a function of temperature for various values of @xmath28 .
when the temperature increases , the zba is suppressed , a behavior similar to what is obtained within the @xmath36 theory.@xcite again , all these curves coincide when one considers their variation with respect to @xmath31 ( see right panel of fig .
[ figure4 ] ) .
[ fig_zero_noise_2 ] shows that the current and the zf noise increase both monotonously with @xmath28 , and that the noise is larger that the current due to thermal fluctuations . in the intermediate temperature regime @xmath128 ,
the zf noise has a totally different dependence on @xmath28 in comparison to its behavior at low temperature due again to an increasing contribution of the thermal noise . however , the total noise is not a direct superposition of shot noise and thermal noise , but results from a more complicated interplay between them.@xcite , for @xmath110 , and @xmath129 .
the arrows indicate the associated values of @xmath111 .
[ fig_zero_noise_2],width=226 ] _ high temperature behavior_. in the equilibrium regime ( i.e. , for @xmath81 ) , the dc current of eq .
( [ eq_current ] ) becomes strictly linear in @xmath17 , and the linear conductance takes the value:@xcite @xmath130~,\end{aligned}\ ] ] where @xmath131 , and @xmath132 is the gamma function . in particular , in the sbs , at @xmath133 , one recovers the power law : @xmath134 this result is in accordance with the generic behavior of the tll in the sbs regime,@xcite where one has @xmath135 , as @xmath2 here .
it is also similar to that obtained within the @xmath36 theory in the tunneling regime , even though the conductor is well - transmitting here.@xcite notice that this dependence on temperatures @xmath136 in eq .
( [ gsbs ] ) is similar to that of the non - linear conductance on voltages @xmath86 , as one can see from differentiating eq .
( [ isbs ] ) .
this confirms , as generally expected , that temperature and voltage play symmetric roles .
more precisely , the differential conductance at both finite temperatures and voltages obeys a scaling law : @xmath137 , where @xmath138 and @xmath139 constant value .
this leads in particular to the same power law behavior with respect to @xmath140 . while this is valid in the sbs regime as we have just shown , with @xmath141 , it turns out that this scaling behavior is violated in the opposite limit of wbs as we discuss in detail now .
more precisely , one has to consider the backscattering conductance , defined as the differential of the backscattering current in eq .
( [ back ] ) : @xmath142 . using eq .
( [ diff_cond_v_0 ] ) at @xmath136 yields : @xmath143 in accordance with the behavior @xmath144 at arbitrary @xmath11 .
we see , however , that this dependence on temperature is different from that on voltage obtained by expanding eq .
( [ diff_cond_t_0 ] ) in the wbs regime ( @xmath89 ) ( see also eq .
( [ backexpansion ] ) ) : @xmath145 as we have already commented concerning eq .
( [ backexpansion ] ) , for generic values of @xmath11 , the lowest order expansion of the backscattering current is given by @xmath146 , leading to a differential conductance proportional to @xmath147 , which cancels at @xmath2 .
this can be seen as well by differentiating the constant term on the r.h.s .
( [ backexpansion ] ) : thus one needs the next term , yielding to the first non - zero contribution in eq .
( [ gb ] ) .
in this section , we discuss the dynamical properties of our system by performing plot analysis of the formal expressions of the ff conductance @xmath25 $ ] and the ff non - symmetrized noise @xmath148 obtained in sec .
both quantities depend not only of the voltage @xmath17 and frequency @xmath149 , but also on temperature .
more precisely , they can be cast into a scaling form , being a function of @xmath30 , @xmath32 , and @xmath31 .
all the dependencies can be obtained exactly , provided these energies are below @xmath78 , but we have to make restrictions on the number of curves presented . here
we will focus first on low temperatures compared to voltage and frequency , @xmath150 , thus both out - of - equilibrium and quantum regime , allowing for @xmath151 to be of the order or greater than @xmath3 .
then we will consider intermediate and high temperatures compared to the voltage .
_ low temperature behavior_. at @xmath152 , using eqs .
( [ eq_g_i ] ) and ( [ eq_current_t_0 ] ) , the ff conductance reads as : @xmath153=\frac{g_q}{2}\left[1-\frac{ev_b}{4\hbar\omega}\sum_\pm\mathrm{arctan}\left(\frac{2\hbar\omega\pm e v}{ev_b}\right)\right]~.\nonumber\\\end{aligned}\ ] ] for @xmath47 , it reduces to the perfect conductance
equal to @xmath105 whereas for @xmath85 , it acquires a frequency dependence as shown on fig .
[ figure6 ] .
since @xmath25 $ ] is an even function of @xmath149 , it is plotted at positive frequencies only .
this result is interesting in the sense that the ff conductance has a non monotonous behavior .
it exhibits a minimum ( see the left panel of fig .
[ figure6 ] ) at a frequency depending on both @xmath17 and @xmath29 which is fixed by the cancellation of the derivative @xmath154 $ ] .
when @xmath28 decreases , this non monotonous behavior disappears , and the ff conductance increases regularly with frequency .
all these curves scale to the same one when one considers their variation with respect to @xmath32 , at fixed @xmath30 and @xmath31 ( see right panel of fig .
[ figure6 ] ) .
notice that because of the finite value of the voltage , the ff conductance acquires a non - zero value even though small at zero frequency .
this has been checked by zooming around @xmath155 ( not shown ) .
, as a function of @xmath156 , for different values of @xmath28 , at @xmath110 and @xmath95 .
right panel : all the curves of the left graphic scale to a single one when one considers their variation with @xmath32 .
we take @xmath157 and @xmath101 .
[ figure6],title="fig:",width=158 ] , as a function of @xmath156 , for different values of @xmath28 , at @xmath110 and @xmath95 .
right panel : all the curves of the left graphic scale to a single one when one considers their variation with @xmath32 .
we take @xmath157 and @xmath101 .
[ figure6],title="fig:",width=158 ] we turn now our attention to the ff noise . at @xmath152 , the integral in eq .
( [ eq_noise ] ) can be performed analytically and the ff non - symmetrized noise reads : @xmath158 where the dimensionless function @xmath159 is given by : @xmath160\mathrm{arctan}\left({2\tilde\omega\pm \tilde v}\right)\bigg]\nonumber\\ & & + \frac{1}{8\tilde\omega}\sum_{\pm}\big[-\theta(-\tilde\omega)+\theta(-\tilde\omega\pm \tilde v)\big]\nonumber\\ & & \times\bigg[\mathrm{ln}\left(1+{\tilde v^2}\right)-\mathrm{ln}\left(1+{(2\tilde\omega\mp \tilde v)^2}\right)\bigg]~.\end{aligned}\ ] ] here @xmath161 is the heaviside function , @xmath162 and @xmath163 .
notice that the ff noise is an even function of @xmath17 .
because of this parity , we will specify only to positive values of the voltage .
we recall that the ff non - symmetrized noise at positive ( negative ) frequencies corresponds to emission ( absorption )
noise.@xcite an important fact we observe is that the emitted noise vanishes exactly above @xmath3 . in the perturbative regime
, one usually expects that the emitted noise vanishes above @xmath164.@xcite indeed , it has been shown universally for a weak tunneling junction ( between strongly correlated systems in arbitrary dimensions and with possible coupling to an environment ) that the ff noise associated to tunneling vanishes above @xmath165 , where @xmath166 is the transferred charge .
@xcite nevertheless , when higher order processes with respect to tunneling are taken into account , one expects the implication of a many - body complicated process , and energy conservation at the level of one particle can not be used any more . in strongly correlated systems or in a conductor coupled to an environment , we are not aware of any non - perturbative statement about the existence of a value of the frequency @xmath149 over which the emitted noise vanishes . however , we obtain an exact result here : the emitted noise vanishes strictly at frequencies @xmath167 , whatever the values of @xmath17 and @xmath29 are .
this is due to the bose - einstein distribution functions of eq .
( [ eq_noise_bis ] ) which reduce , in the zero temperature limit , to the heaviside functions of eq .
( [ eq_fct_f ] ) . their presence
is related to the photon exchange between the conductor and the environment corresponding to electron - hole type excitations .
one can think of two possible scenarios : ( i ) the new fermions are independent . for non - interacting electrons
, scattering theory predicts that the emitted noise vanishes above @xmath3 . even though the ff noise in our case does not obey the same relation with respect to the transmission amplitude @xmath22
, one can imagine that a similar fact holds .
( ii ) the second scheme is that the system can not emit at higher frequencies compared to what the generator can afford , and not compared to the voltage across the conductor @xmath65 if it was perfect .
if this is plausible , it opens the question as to whether such a result is universal for the tll with an arbitrary value of the parameter @xmath11 , thus for other values of the resistance @xmath6 .
equivalently , for @xmath168 , we have @xmath169 , and eq .
( [ def_g ] ) reduces to:@xcite @xmath170~,\end{aligned}\ ] ] which means that at negative frequencies below @xmath171 and zero temperature , the absorption ff noise is governed by the ff conductance .
it is also interesting to express the ff non - symmetrized excess noise which is often measured experimentally , both because zero - point fluctuations are not easy to measure ( apart from a pioneering work in ref . ) , and in order to subtract undesirable sources of noise .
it is given by : @xmath172 at zero temperature , it reads : @xmath173~ , \end{aligned}\ ] ] where the function @xmath159 is given by eq .
( [ eq_fct_f ] ) .
, as a function of @xmath156 , for different values of @xmath28 , at @xmath110 and @xmath95 .
right panel : non - symmetrized excess noise at negative frequency , in units of @xmath174 , as a function of @xmath32 , at @xmath157 and @xmath101.[figure7],title="fig:",width=158 ] , as a function of @xmath156 , for different values of @xmath28 , at @xmath110 and @xmath95 .
right panel : non - symmetrized excess noise at negative frequency , in units of @xmath174 , as a function of @xmath32 , at @xmath157 and @xmath101.[figure7],title="fig:",width=158 ] , as a function of @xmath156 , for different values of @xmath28 , at @xmath110 and @xmath95 .
right panel : non - symmetrized excess noise at positive frequency , in units of @xmath174 , as a function of @xmath32 , at @xmath157 and @xmath101.[figure8],title="fig:",width=158 ] , as a function of @xmath156 , for different values of @xmath28 , at @xmath110 and @xmath95 .
right panel : non - symmetrized excess noise at positive frequency , in units of @xmath174 , as a function of @xmath32 , at @xmath157 and @xmath101.[figure8],title="fig:",width=158 ] in the left panels of figs .
[ figure7 ] and [ figure8 ] is plotted the excess noise at a very low temperature@xcite at positive frequency and negative frequency respectively .
the red curve corresponds to a value of @xmath121 , i.e. to the wbs regime . in that case one
sees , according to eq .
( [ eq_fct_f ] ) , that the dependence close to @xmath175 is due mainly to the @xmath176 function : this explains the step whose width is controlled by @xmath29 .
the step is smoothed out at higher @xmath29 ( i.e. , lower @xmath28 ) in the green and blue dashed curves of fig . [ figure7 ] .
however , the asymmetry between absorbed and emitted excess noise is always present whatever the value of the effective transmission is .
this asymmetry has been shown to be related , in a universal way , to non - linearity , using eq .
( [ def_g]).@xcite here , the non - linearity is induced by the electromagnetic environment .
all these curves at different @xmath28 scale to a unique one as it is shown in the right panels of figs .
[ figure7 ] and [ figure8 ] where @xmath177 is plotted as a function of @xmath32 when the ratios @xmath30 and @xmath31 are fixed .
we have to draw attention to the fact that this requires us to change simultaneously all energy scales while @xmath29 changes .
, as a function of @xmath178 , for different values of @xmath28 , at @xmath127 and @xmath179 .
right panel : all the curves of the left graphic scale to a single one when one considers their variations with respect to @xmath31 .
we take @xmath127 and @xmath180.[figure9],title="fig:",width=158 ] , as a function of @xmath178 , for different values of @xmath28 , at @xmath127 and @xmath179 .
right panel : all the curves of the left graphic scale to a single one when one considers their variations with respect to @xmath31 .
we take @xmath127 and @xmath180.[figure9],title="fig:",width=158 ] , as a function of @xmath156 , for different values of @xmath28 .
we use the values @xmath110 , and @xmath129 .
right panel : non - symmetrized excess noise , in units of @xmath174 , as a function of @xmath32 , at @xmath157 and @xmath181.[figure10],title="fig:",width=158 ] , as a function of @xmath156 , for different values of @xmath28 .
we use the values @xmath110 , and @xmath129 .
right panel : non - symmetrized excess noise , in units of @xmath174 , as a function of @xmath32 , at @xmath157 and @xmath181.[figure10],title="fig:",width=158 ] _ intermediate temperature behavior_. next , we look at the effect of the temperature on the ff conductance and the ff noise .
the left panel of fig .
[ figure9 ] shows that the ff conductance increases monotonously with temperature .
again , all the curves scale to a single one when one considers the variation with respect to @xmath31 at fixed @xmath32 ( see the right panel of fig . [ figure9 ] ) .
the ff excess noise at intermediate temperatures is plotted on the left panel of fig .
[ figure10 ] .
however , the asymmetry of the ff excess noise is still visible in that regime , and we observe its enhancement when the effective transmission decreases . again
, all the curves scale to a single one when one considers the variation with @xmath32 at fixed values of @xmath30 and @xmath31 ( see the right panel of fig .
[ figure10 ] ) .
_ high temperature behavior_. at high temperatures compared to voltage , @xmath81 , we have checked , using eqs .
( [ eq_noise_bis ] ) and ( [ eq_g_i ] ) , that the ff non - symmetrized noise is related to the ff conductance and becomes voltage independent in that limit according to the fluctuation - dissipation theorem : @xmath182~.\nonumber\\\end{aligned}\ ] ]
in this paper , we have studied both stationary and time - dependent transport properties of a well - transmitting one - channel conductor embedded in an ohmic environment with a quantum of resistance @xmath0 .
we have taken advantage of the mapping of this problem to a tll with a particular value of the interaction parameter @xmath2 .
this has allowed us to obtain results which are non - perturbative with respect to the resistance of the environment @xmath6 ( here being equal to @xmath183 , see eq .
( [ key ] ) ) , and which describes all regimes of voltages , frequencies and temperatures below the cutoff @xmath27 .
the results are controlled by a unique scaling and non - universal parameter @xmath29 which depends on both the properties of the conductor and the environment .
while the dc properties can as well be derived exactly for other values of @xmath6 ( analytically at zero temperature , otherwise one would need numerical methods ) , using the bethe - ansatz solution for the impurity problem in the tll , the particular choice of @xmath12 has allowed us to obtain the first non - perturbative results for both the ff noise and the ff conductance . for the latter ,
we have used its universal relation to the asymmetry of the emission and absorption noise , via an out - of - equilibrium fluctuation - dissipation theorem type formula ( see eq .
( [ def_g ] ) ) derived in ref . .
we have first recalled the results for the dc transport obtained in a tll with parameter @xmath2 in order to apply them to our present problem and to discuss them in more details .
the key procedure is based on refermionization , i.e. , casting the non - trivial strong correlation effects into new chiral fermions with an energy - dependent transmission coefficient @xmath184 .
this can not be obtained by any `` adiabatic '' evolution from the effective transmission @xmath28 ( which explains our choice for a different notation ) .
the dc current through the mesoscopic conductor takes a form similar to that within the scattering theory , applied to the new fermions : it is expressed as the integral over energy of @xmath184 times the difference of the fermi - dirac distribution functions of the left and right reservoirs . even for a good effective transmission @xmath28 ,
one recovers the dcb at energies below the voltage @xmath29 .
the reduction to the dc current has a power law with different exponents when voltages are high or low compared to @xmath29 , corresponding to the wbs and sbs regimes .
in particular , in the latter limit , one recovers the behavior predicted within the @xmath36 theory for @xmath86 for a weakly transmitting conductor , even though we consider a well transmitting one here . in these two opposite limits ,
the zf noise becomes poissonian , with a fano factor respectively given by @xmath77 or @xmath120 .
the former renormalization is related to the conductance of the conductor in the perfect limit , being in series with the environmental quantum resistance .
the zf noise obeys as well the same expression in terms of @xmath184 as within the scattering approach .
nevertheless , this does not hold for the ff noise even though it can be expressed in terms of the transmission amplitude @xmath22 : this is a surprising and crucial result of our paper . another
interesting fact is that the emitted ff noise vanishes strictly above @xmath3 at zero temperature , a fact common to non - interacting electrons , which is not obvious to expect within the underlying strongly correlated system .
we have shown that the ff conductance does not obey the scattering approach formulation for the new fermions .
rather , it obeys a simple relation that involves dc currents .
this kind of relation was previously shown for tunneling barriers within the tien - gordon theory , or in the fqhe at simple fillings.@xcite recently it has been extended to arbitrary filling factors , and even more has been shown to be universal either for tunneling barriers in arbitrary dimensions , as well as weak barriers in one dimension.@xcite it is quite remarkable that such a relation extends to arbitrary regimes within the tll model at @xmath2 .
it turns out that the ff conductance has a non - monotonous behavior with respect to frequency , having a minimum at a frequency that depends both on the applied voltage and the scaling voltage @xmath29 , and the value of which is reduced when the effective transmission @xmath28 decreases .
we have shown that both the ff conductance and the ff noise have a universal behavior at different @xmath28 when voltages , temperatures and frequencies are all divided by the same scaling voltage @xmath29 .
this extends the result obtained for the differential conductance , valid for arbitrary values of the parameter @xmath11 thus of the resistance @xmath6 in our problem where we expect to get the same scaling behavior for time - dependent transport as well .
the coherent conductor connected to an ohmic environment offers a unique framework to realize a tll with a tunable parameter @xmath11 , and a unique possibility to realize @xmath2 .
the present experiments on that direction are very promising.@xcite beyond this issue , our work provides benchmark results for time - dependent transport in various fundamental problems : the dcb phenomena , other strongly correlated systems where the special value @xmath2 has been studied fully,@xcite and more generally those where one solves the interacting problem in terms of new independent particles , such as within the bethe - ansatz methods for the impurity problem in tll.@xcite we have highlighted novel counter - intuitive facts , in particular , we have shown that one can not systematically apply the scattering approach to those new independent particles .
i.s . is grateful to a. anthore , s. jzouin , f.d .
parmentier , f. pierre for their insightful remarks and to s. carr , b. douot , d. estve , c. glattli , p. joyez , h. saleur , p. simon , j.r .
souquet , e. sukhorukov and a. zaikin for fruitful discussions ; she acknowledges financial support of anr under contract anr-09-blan-0199 - 01 .
thank t. martin for preliminary discussions about the refermionization procedure .
the mapping of ref . applies to the effective action once degrees of freedom apart from those at @xmath185 are integrated out .
it is more convenient , for the present computation , to consider the extended tll hamiltonian over space coordinate @xmath186 , with an impurity located at @xmath185 ( see fig .
[ figure11 ] ) : @xmath187\nonumber\\ & & + \sqrt{\frac{\hbar\omega_fev_b}{32\pi^3}}e^{i\phi(0,t)-ievt/2\hbar}+h.c.~,\end{aligned}\ ] ] where the bosonic fields @xmath188 and @xmath189 are related to the initial bosonic fields describing right ( r ) and left ( l ) movers through : @xmath190 where @xmath191 are the initial fermionic operators .
@xmath192 is the distance cutoff of the tll theory .
the klein factors @xmath193 allow to have the proper commutation relation for the fermionic operators .
notice that in eq .
( [ hamilton ] ) , the bosonic field refers to @xmath194 .
the average current through the conductor is given by:@xcite @xmath195 where @xmath196 is the fermi velocity .
the operators @xmath197 and @xmath198 refer to the densities of right and left moving electrons in tll reservoirs ( see fig .
[ figure11 ] ) .
they are related to the new fermionic operators introduced in the refermionization procedure through the relations : @xmath199 where @xmath200 and @xmath201 are the new fermionic fields associated to the bosonic fields @xmath188 and @xmath189 : @xmath202 is affected by backscattering and by the applied voltage ( see eq .
( [ h_annexe ] ) ) , whereas @xmath203 is neither affected by backscattering , nor by the applied voltage . in eq .
( [ a1 ] ) , the average value is defined as the average over the scattering state@xcite minus the average at equilibrium ( i.e. , @xmath127 ) , since we have to take the normal order@xcite in the products of operators that appear in eqs .
( [ a2 ] ) and ( [ a3 ] ) .
thus , we immediately conclude that @xmath204 .
the klein factors @xmath205 and @xmath206 allow us to have the proper commutation relation for the new fermionic operators . is the energy scale that characterizes backscattering of electrons at the conductor position @xmath185.[figure11],width=302 ] from eqs .
( [ a2 ] ) and ( [ a3 ] ) , we understand that @xmath207 and @xmath208 depend on the position @xmath186 , however , the average of the difference @xmath209 , which appears in the current , does not depend on the position , since the average of the total current is a conserved quantity ( it is the reason why we do not keep the @xmath186 dependency in the current ) . using this hamiltonian , one can write the equations of motion for the fields .
the solution for the fermionic field @xmath210 , associated with the free bosonic field , is that of free propagating electrons , whereas the solution for @xmath211 is given by@xcite ( from now , we take @xmath212 ) : @xmath213 where the operator @xmath214 is a combination of the annihilation and creation operators , @xmath215 and @xmath216 , which obey the commutation relation @xmath217 .
the transmission and reflection amplitudes read as : @xmath218 with the help of the solutions given by eqs .
( [ sol1 ] ) and ( [ sol2 ] ) , one can calculate the average over the product of the two fermionic fields : @xmath219~,\end{aligned}\ ] ] and , @xmath220~,\end{aligned}\ ] ] where @xmath221 .
we have used the following average values : @xmath222 , and @xmath223 , with @xmath68 is the fermi - dirac distribution function . reporting eqs .
( [ av1 ] ) and ( [ av2 ] ) into eqs .
( [ a1 ] ) and ( [ a2 ] ) , we finally obtain eq .
( [ eq_current ] ) .
to calculate the ff non - symmetrized noise , we need to evaluate the following correlators ( @xmath212 ) : @xmath224 the determination of @xmath225 is rather simple since it involves contributions which are not affected by the backscattering : @xmath226 we use wick s theorem to calculate the correlator @xmath227 . after successive integrations over frequencies and times ,
we obtain : @xmath228~.\end{aligned}\ ] ] the calculation of @xmath229 and @xmath230 mix @xmath216 and @xmath231 operators since : @xmath232 and @xmath233 we obtain @xmath234 with : @xmath235\nonumber\\ & & \times \big[t(\omega')t^*(\omega')t(\omega+\omega')t^*(\omega+\omega')\nonumber\\ & & -r(\omega')r^*(\omega')r(\omega+\omega')r^*(\omega+\omega')\big]~.\end{aligned}\ ] ] next , we calculate @xmath236 which involves @xmath231 and @xmath237 operators since : @xmath238 applying wick s theorem , we obtain : @xmath239\nonumber\\ & & \times\big[|t(\omega')|^2|t(\omega+\omega')|^2+|r(\omega')|^2|r(\omega+\omega')|^2\nonumber\\ & & -2t(\omega')t^*(\omega+\omega')r^*(\omega')r(\omega+\omega')\big]\nonumber\\ & & + \sum_\pm f\left(\hbar\omega+\hbar\omega'\pm ev/2\right)\left(1-f\left(\hbar\omega'\mp ev/2\right)\right)\nonumber\\ & & \times \big[|t(\omega')|^2|r(\omega+\omega')|^2\nonumber\\ & & + t(\omega')t^*(\omega+\omega')r^*(\omega')r(\omega+\omega')\big]\bigg\}~.\end{aligned}\ ] ] finally , the ff non - symmetrized noise of eq .
( [ eq_noise ] ) is obtained from @xmath240/4 $ ] , where we replace , everywhere it appears , @xmath241 by @xmath242 .
in this appendix , we compare the expressions of current , ff conductance and noise that we have established to those obtained in the framework of the scattering theory . it has been already noted that the expressions of the dc current given by eq .
( [ eq_current ] ) and the zf noise are formally identical to those derived from the scattering approach ( i.e. , they are expressed as integrals of the transmission coefficient times fermi dirac distribution functions ) , even though the non - trivial many - body effects are encoded into this transmission .
we show here that the ff conductance and the ff noise , even though expressed in terms of the transmission and reflection amplitudes , do not obey the relation derived within the scattering approach .
@xmath243\big[f_{++}(\omega,\omega')+f_{--}(\omega,\omega')\big]\nonumber\\ + \big[\mathcal{t}(\omega')-\mathcal{t}(\omega')\mathcal{t}(\omega+\omega')-|t(\omega')-t(\omega+\omega')|^2/4\big]\big[f_{+-}(\omega,\omega')+f_{-+}(\omega,\omega')\big]\bigg)~,\end{aligned}\ ] ] where @xmath244 $ ] . using our notations , the symmetrized noise of a coherent one - channel coherent conductor with an energy - dependent transmission @xmath73 can be obtained within the scattering approach:@xcite @xmath245f_{++}(\omega,\omega')\nonumber\\ + \mathcal{t}(\omega')\mathcal{t}(\omega+\omega')f_{--}(\omega,\omega')+\mathcal{t}(\omega')(1-\mathcal{t}(\omega+\omega'))f_{+-}(\omega,\omega')+ \mathcal{t}(\omega+\omega')(1-\mathcal{t}(\omega'))f_{-+}(\omega,\omega')\bigg)~.\end{aligned}\ ] ] the main difference between eqs .
( [ eq_sym_noise ] ) and ( [ eq_buttiker ] ) resides into the factors in front of @xmath246 and @xmath247 which are identical in eq .
( [ eq_sym_noise ] ) but not in eq .
( [ eq_buttiker ] ) .
the same remark applies to the factors in front of @xmath248 and @xmath249 .
if one had a transmission coefficient @xmath73 independent on energy , both expressions would have been identical .
20 yu.v .
nazarov , zhetf * 95 * , 975 ( 1989 ) [ sov .
jetp * 68 * , 561 ( 1989 ) ] ; p. delsing , k.k .
likharev , l.s .
kuzmin , and t. claeson , phys .
lett . * 63 * , 1180 ( 1989 ) ; g. schn and a.d .
zaikin , phys . rep . * 198 * , 237 ( 1990 ) ; m.h .
devoret , d. estve , h. grabert , g .-
ingold , h. pothier , and c. urbina , phys .
64 * , 1824 ( 1990 ) ; g .-
ingold and yu.v .
nazarov , in _
charge tunneling rates in ultrasmall junctions _ , _ b294 _ , h. grabert and m.h .
devoret eds .
( plenum press , new york , 1992 ) , chap . 2 , p. 21 .
the case of many weakly transmitting channels with a total large conductance was treated by p. joyez , d. estve , and m.h .
devoret , phys .
lett . * 80 * , 1956 ( 1998 ) .
golubev and a.d .
zaikin , phys .
lett . * 86 * , 4887 ( 2001 ) .
a. levy yeyati , a. martin - rodero , d. estve , and c. urbina , phys .
lett . * 87 * , 046802 ( 2001 ) . c. altimiras , u. gennser , a. cavanna , d. mailly , and f. pierre , phys .
99 * , 256805 ( 2007 ) .
m. kindermann and yu.v .
nazarov , phys .
lett . * 91 * , 136802 ( 2003 ) . i. safi and h. saleur , phys
. rev . lett . * 93 * , 126602 ( 2004 ) .
s. tomonaga , prog .
phys . * 5 * , 544 ( 1950 ) .
luttinger , j. math .
phys . * 4 * , 1154 ( 1963 ) .
we use the notation @xmath10 for the energy that characterizes the crossover between the wbs and sbs regimes instead of the notation @xmath250 used in ref . .
f.d . parmentier , a. anthore , s. jezouin , h. le sueur , u. gennser , a. cavanna , d. mailly , and f. pierre , nature physics * 7 * , 935 ( 2011 ) .
golubev , a.v .
galaktionov , and a.d .
zaikin , phys .
b * 72 * , 205417 ( 2005 ) .
p. fendley , a.w.w .
ludwig , and h. saleur , phys .
lett . * 74 * , 3005 ( 1995 ) .
i. safi and h.j .
schulz , edited by t. martin , g. montambaux , and j.t.t .
van , eds frontires , gif - sur - yvette ( 1996 ) .
a. furusaki and n. nagaosa , phys .
b * 54 * , r5239 ( 1996 ) .
f. dolcini , b. trauzettel , i. safi , and h. grabert , phys .
b * 71 * , 165309 ( 2005 ) .
lebedev , a. crpieux , and t. martin , phys .
b * 71 * , 075416 ( 2005 ) .
m. guigou , a. popoff , t. martin , and a. crpieux , phys .
b * 76 * , 045104 ( 2007 ) . c. de c. chamon , d.e
. freed , and x.g .
wen , phys .
b * 53 * , 4033 ( 1996 ) .
f. lesage and h. saleur , nucl .
b * 490 * , 543 ( 1997 ) .
notice that the effective charge @xmath251 corresponds here to @xmath252 , but does not correspond to any physical entity within the dcb problem : it is related simply to the conductance in series of a perfect conductor , with a quantum of resistance , and the resistance @xmath6 .
r. aguado and l p. kouwenhoven , phys .
lett . * 84 * , 1986 ( 2000 ) ; r. deblock , e. onac , l. gurevich , and l.p .
kouwenhoven , science * 301 * , 203 ( 2003 ) ; p .- m .
billangeon , f. pierre , h. bouchiat , and r. deblock , phys .
* 96 * , 136804 ( 2006 ) ; e. onac , f. balestro , l.h .
willems van beveren , u. hartmann , y.v .
nazarov , and l. p. kouwenhoven , _ ibid . _
* 96 * , 176601 ( 2006 ) ; s. gustavsson , m. studer , r. leturcq , t. ihn , k. ensslin , d.c .
driscoll , and a.c .
gossard , _ ibid . _ * 99 * , 206804 ( 2007 ) ; e. zakka - bajjani , j. sgala , f. portier , p. roche , d.c .
glattli , a. cavanna , and y. jin , _ ibid . _ * 99 * , 236803 ( 2007 ) . c. bena and i. safi , phys .
b * 76 * , 125317 ( 2007 ) .
the symmetrized noise one can deduce from such a computation fits perfectly with that obtained through a different approach by golubev _
et al._,@xcite again perturbatively with respect to @xmath37 .
d. yue , l.i .
glazman , and k.a .
matveev , phys .
b * 49 * , 1966 ( 1994 ) .
m. kindermann and b. trauzettel , phys .
lett . * 94 * , 166803 ( 2005 ) .
i. safi , c. bena , and a. crpieux , phys .
b * 78 * , 205422 ( 2008 ) .
moca , p. simon , c.h .
chung , and g. zarand , phys .
b * 83 * , 201303(r ) ( 2011 ) .
i. safi , arxiv:0908.4382 ( 2009 ) ; aip conf
* 1129 * , 431 , noise and fluctuations : 20th international conference on noise and fluctuations , vol .
1129 ( aip , melville , new york , 2009 ) , eds .
m. macucci and g. basso , cond - mat/0906.4658 . for antennas , or 1d systems with a specific kinetic hamiltonian ,
see also : u. gavish , y. imry , and b. yurke , proceedings of the spie noise conference , grand canay ( 2004 ) , arxiv : cond - mat/0404270 ; u. gavish , phd thesis , weizmann institute of science ( 2003 ) .
i. safi and p. joyez , phys .
b * 84 * , 205129 ( 2011 ) . p.k . tien and j.p .
gordon , phys . rev . * 129 * , 647 ( 1963 ) ; j. bardeen , phys .
* 6 * , 57 ( 1961 ) ; m.h .
cohen , l.m .
falicov , and j.c .
phillips , phys .
* 8 * , 316 ( 1962 ) ; a.h .
worsham , n.g .
ugras , d. winkler , d.e .
prober , n.r .
erickson , and p.f .
goldsmith , phys .
lett . * 67 * , 3034 ( 1991 ) .
tucker , ieee j. quan .
electron . , * 15 * , 1234 ( 1979 ) ; j.r .
tucker and m.j .
feldman , rev .
57 * , 1055 ( 1985 ) .
i. safi _ et al _ , in preparation .
i. safi and e. sukhorukov , eur .
91 * , 67008 ( 2010 ) .
f. guinea , phys .
b * 32 * , 7518 ( 1985 ) .
u. weiss , m. sassetti , t. negele , and m. wollensak , z. phys .
b * 84 * , 471 ( 1991 ) .
c.l . kane and m.p.a .
fisher , phys . rev .
b * 46 * , 15233 ( 1992 ) . p.
fendley , a.w.w .
ludwig , and h. saleur , phys .
b * 52 * , 8934 ( 1995 ) . in general , in the tll model , the cutoff procedure can contain more than one cutoff energy , a tricky problem which we do not address fully here .
the mapping requires to cut the frequencies in the action at @xmath27 , thus the bosonic correlation functions are functions of energy divided by @xmath78 .
thus we adopt this unique cutoff throughout the paper .
i. safi , private communication .
r. egger and h. grabert , phys .
b * 58 * , 10761 ( 1998 )
. c.l . kane and m.p.a .
fisher , phys .
lett . * 68 * , 1220 ( 1992 ) .
see also : t. giamarchi and h.j .
schulz , phys .
b * 37 * , 325 ( 1988 ) where the most complicated situation of an extended disorder had been studied .
h. saleur and u. weiss , phys .
b * 63 * , 201302(r ) ( 2001 ) .
a. komnik and h. saleur , phys .
lett . * 96 * , 216406 ( 2006 ) .
a. komnik , b. trauzettel , and u. weiss , ann .
( leipzig ) * 16 * , 661 ( 2007 ) .
a. herzog and u. weiss , new j. phys . * 10 * , 045008 ( 2008 ) . s. maier and a. komnik , phys .
b * 82 * , 165116 ( 2010 ) ; l. mhlbacher , d.f .
urban , and a. komnik , phys .
b * 83 * , 075107 ( 2011 ) .
the low transmitting regime , @xmath253 , is not a good starting point for the theory . indeed ,
in this limit , marginal process can dominate over the irrelevant tunneling process , and one can not obtain universal results . this is opposite to what happens in the good transmitting regime where the backscattering hamiltonian in eq .
( [ hamilton ] ) is relevant in the rg sense . k.a .
matveev , phys .
b * 51 * , 1743 ( 1995 ) .
a. furusaki , phys .
b * 56 * , 9352 ( 1997 ) .
j. von delft , g. zarand , and m. fabrizio , phys .
lett . * 81 * , 196 ( 1998 ) ; j. von delft , h. schoeller , ann . phys . * 7 * , 225 ( 1998 ) .
d.n . aristov and p. wlfle ,
b * 80 * , 045109 ( 2009 ) . from eq .
( [ eq_current ] ) , we directly see that the current is an odd function with voltage . ya.m .
blanter and m. bttiker , phys . rep . *
336 * , 1 ( 2000 ) . t. martin and r. landauer , phys .
b * 45 * , 1742 ( 1992 ) .
m. bttiker , phys .
b * 45 * , 3807 ( 1992 ) . s .-
eric yang , solid stat . com .
* 81 * , 375 ( 1992 ) .
i.p . levkivskyi and e.v .
sukhorukov , phys .
b * 85 * , 075309 ( 2012 ) . from this relation , one can check easily that the real part of the ff conductance @xmath25 $ ] is an even function with voltage .
f. pierre _
et al . _ ,
private communication . at strictly zero - temperature , the integral in eq .
( [ eq_g ] ) can be performed analytically and the derivative of the differential conductance according to @xmath28 can be obtained from eq .
( [ diff_cond_t_0 ] ) , it reads : @xmath254 . since @xmath255 , we have @xmath256 which can be quite large for @xmath257 . s.m .
girvin , l.i .
glazman , m. jonson , d.r .
penn , and m.d .
stiles , phys .
64 * , 3183 ( 1990 ) .
billangeon , f. pierre , h. bouchiat , and r. deblock , phys .
* 96 * , 136804 ( 2006 ) .
j. basset , h. bouchiat , and r. deblock , phys .
lett . * 105 * , 166801 ( 2010 ) . in figs .
[ figure7 ] and [ figure8 ] , the curves are plotted at very low but small temperature in order to avoid artificial discontinuities . | we consider a one - channel coherent conductor with a good transmission embedded into an ohmic environment , impedance of which is equal to the quantum of resistance @xmath0 below the @xmath1 frequency .
this choice is motivated by the mapping of this problem to a tomonaga - luttinger liquid with one impurity , the interaction parameter of witch corresponds to the specific value @xmath2 , allowing for a refermionization procedure .
the `` new '' fermions have an energy - dependent transmission amplitude which incorporates the strong correlation effects and yields the exact dc current and zero - frequency noise through expressions similar to those of the scattering approach .
we recall and discuss these results for our present purpose .
then we compute , for the first time , the finite - frequency differential conductance and the finite - frequency non - symmetrized noise .
contrary to intuitive expectation , both can not be expressed within the scattering approach for the new fermions , even though they are still determined by the transmission amplitude .
even more , the finite - frequency conductance obeys an exact relation in terms of the dc current which is similar to that derived perturbatively with respect to weak tunneling within the tien - gordon theory , and extended recently to arbitrary strongly interacting systems coupled eventually to an environment or / and with a fractional charge .
we also show that the emission excess noise vanishes exactly above @xmath3 , even though the underlying tomonaga - luttinger liquid model corresponds to a many - body correlated system .
our results apply for all ranges of temperatures , voltages , and frequencies below the @xmath1 frequency , and they allow us to explore fully the quantum regime . |
in @xcite we investigated the theory of _ reticular legendrian unfoldings _ which describes bifurcations of wavefronts on a hypersurface germ with a boundary , a corner , or an @xmath0-corner in a smooth manifold and studied the stabilities and a generic classification of such bifurcations . in this paper
we give a semi - local classification of bifurcations of wavefronts generated by a hypersurface with an @xmath0-corner .
the wavefronts generated around several points on the initial hypersurface intersect .
for example we consider the below figures in which the initial wavefront with a boundary in a plane and wavefronts generated by the initial wavefront and the boundary to normal directions are described respectively
. with a boundary and generated wavefronts ( @xmath1 , @xmath2),width=566,height=170 ] the generated wavefronts bifurcate and typical shapes of bifurcations occur .
the @xmath3 , @xmath4 , and @xmath5 fronts are intersections of fronts generated by different points of the initial wavefront .
the purpose of this paper is the classification of such generic bifurcations of intersections of wavefronts . as the bifurcation theory of wavefronts
are described by reticular legendrian unfoldings , we shall introduce the notion of _ multi - reticular legendrian unfoldings _ in order to describe the theory of generic bifurcations of intersections of wavefronts generated by different points of a hypersurface with an @xmath0-corner .
the main result of this paper is a classification of generic bifurcations of intersections of wavefronts generated by a hypersurface without boundary or with boundaries in a manifold with dimension @xmath6 and @xmath7 .
the classification list of generating families of generic multi - reticular legendrian unfoldings are given before theorem [ genericclass ] .
we also draw all figures of such bifurcations at the last of this paper .
this paper consists of four sections . in section [
unfold : sec ] we investigate the theory of map germs with respect to _ the reticular @xmath8-@xmath9-@xmath10-equivalence relation _ which plays important roles as generating families of multi - reticular legendrian unfoldings . in section [ retlegunf : sec ] , we review the theory of reticular legendrian unfoldings which is developed in @xcite . in section [ mrlu : sec ]
we introduce the notion of multi - reticular legendrian unfoldings and investigate their equivalence relation , generating families , and stabilities . in section [ generic : sec ]
we reduce our investigation to finitely dimensional jet spaces and give a generic classification of multi - reticular legendrian unfoldings .
in this section we investigate the theory of map germs with respect to _ the reticular @xmath8-@xmath9-@xmath10-equivalence relation _ which is proved by almost parallel methods of @xcite .
we denote by @xmath11 the set of all germs at @xmath12 of smooth maps @xmath13 and set @xmath14 .
we denote @xmath15 for @xmath16 and denote @xmath17 for @xmath18 . if @xmath19 we write simply @xmath20 for @xmath21 and @xmath22 for @xmath23 .
then @xmath20 is an @xmath24-algebra in the usual way and @xmath22 is its unique maximal ideal .
we also denote by @xmath25 for @xmath26 and @xmath27 for @xmath28 .
we remark that @xmath29 is an @xmath20-module generated by @xmath30-elements .. we denote by @xmath31 the set of @xmath32-jets at @xmath12 of germs in @xmath33 .
there are natural projections : @xmath34 we write @xmath35 for @xmath36 for each @xmath37 .
let @xmath38 be a fixed coordinate system of @xmath39 .
we denote by @xmath40 the group of diffeomorphism germs @xmath41 of the form : @xmath42 we denote by @xmath43 the group of diffeomorphism germs @xmath44 of the form : @xmath45 we denote @xmath46 , @xmath47 @xmath48 , and denote other notations analogously . [ gw1.8:cor](cf .
* corollary 1.8 ) ) let @xmath49 be a submodule of @xmath50 , @xmath51 be a finitely generated @xmath52-submodule of @xmath50 generated @xmath53-elements , and @xmath54 be a finitely generated @xmath55 submodule of @xmath50 .
suppose @xmath56 then @xmath57
@xmath58 we say that @xmath59 are _ reticular @xmath60-equivalent _ if there exist @xmath61 and a unit @xmath62 such that @xmath63 for @xmath64 . we denote @xmath65 and we call @xmath66 a reticular @xmath60-isomorphism from @xmath67 to @xmath68
. we remark that @xmath69 and @xmath70 are reticular @xmath60-equivalent if and only if @xmath71 and @xmath72 are reticular @xmath73-equivalent for @xmath64 .
let @xmath74 and @xmath75 .
let @xmath76 be the submanifold of @xmath77 consist of the image by @xmath78 of the orbit under the reticular @xmath60-equivalence of @xmath69 .
put @xmath75 .
then @xmath79 we say that a map germ @xmath80 is _
reticular @xmath60-@xmath32-determined _ if all map germ in @xmath81 which has the same @xmath32-jet of @xmath69 is reticular @xmath60-equivalent to @xmath69 .
[ findetc : lm](cf . ,
* lemma 2.3 ) ) let @xmath80 and let @xmath82 for @xmath64
. then @xmath69 is reticular @xmath60-@xmath32-determined .
conversely if @xmath74 is reticular @xmath60-@xmath32-determined , then @xmath83 we say that @xmath84 are _ reticular @xmath9-@xmath10-equivalent _ if there exist @xmath61 of the form : @xmath85 and a unit @xmath86 such that @xmath87 for @xmath64 .
we write @xmath88 and call @xmath66 a reticular @xmath9-@xmath10-isomorphism from @xmath67 to @xmath68 .
+ we say that a map germ @xmath89 is _
reticular @xmath9-@xmath10-@xmath32-determined _ if all map germ in @xmath90 which has the same @xmath32-jet of @xmath67 is reticular @xmath9-@xmath10-equivalent to @xmath67 .
let @xmath91 and @xmath92 .
we write @xmath93 the submanifold of the image of the orbit of @xmath67 under the reticular @xmath8-@xmath9-@xmath10 equivalence relation . then @xmath94 [ nmqdet](cf . ,
* lemma 3.10 ) ) let @xmath95 and @xmath32 be a non - negative integer . if @xmath96 then @xmath67 is reticular @xmath9-@xmath10-@xmath32-determined . in convenience , we denote an unfolding of a function germ @xmath97 @xmath98 by @xmath99 .
let @xmath100 and @xmath101 be unfoldings of @xmath102 .
_ a reticular @xmath8-@xmath9-@xmath10-@xmath67-morphism from @xmath103 to @xmath104 _ is a pair @xmath105 , where @xmath106 for @xmath107 and @xmath108 is a unit of @xmath109 satisfying the following conditions : + ( 1 ) @xmath110 can be written in the form : @xmath111 ( 2 ) @xmath112 , @xmath113 , + ( 3 ) @xmath114 for all @xmath115 .
if there exists a reticular @xmath8-@xmath9-@xmath10-@xmath67-morphism from @xmath103 to @xmath104 , we say that _ @xmath104 is reticular @xmath8-@xmath9-@xmath10-@xmath67-induced from @xmath103_. if @xmath116 and @xmath117 is invertible , we call @xmath105 _ a reticular @xmath8-@xmath9-@xmath10-@xmath67-isomorphism from @xmath103 to @xmath104 _ and we say that @xmath103 is reticular @xmath8-@xmath9-@xmath10-@xmath67-equivalent to @xmath104 .
we say that @xmath118 are _ reticular @xmath8-@xmath9-@xmath10-equivalent _ if there exist diffeomorphism germs @xmath119 of the form @xmath120 and a unit @xmath121 such that @xmath122 for @xmath64 .
we call @xmath105 _ a reticular @xmath8-@xmath9-@xmath10-isomorphism from @xmath103 to @xmath104_. + we say that function germs @xmath123 @xmath124 and @xmath125 are _ stably reticular @xmath8-@xmath126-@xmath73-equivalent _ if @xmath127 and @xmath128 are reticular @xmath8-@xmath126-@xmath73-equivalent after additions of linear forms of @xmath129 of which all coefficients are not zero and non - degenerate quadratic forms in the variables @xmath130 .
+ we define the notion of _ stable reticular @xmath8-@xmath9-@xmath10-equivalence _ in the same way as the stable reticular @xmath8-@xmath126-@xmath73-equivalence .
we define stabilities of unfoldings .
let @xmath131 @xmath132 be an unfolding of @xmath133 @xmath134 .
+ we say that @xmath103 is _
reticular @xmath8-@xmath9-@xmath10-stable _ if the following condition holds : for any neighbourhood @xmath135 of @xmath12 in @xmath136 and any representative @xmath137 , there exists a neighbourhood @xmath138 of @xmath139 in @xmath140-topology such that for any @xmath141 there exist @xmath142 for @xmath64 such that @xmath143 and @xmath103 are reticular @xmath8-@xmath9-@xmath10-equivalent where @xmath144 is defined by @xmath145 . + we say that @xmath103 is _ a reticular @xmath8-@xmath9-@xmath10-versal unfolding of @xmath67 _ if any unfolding of @xmath67 is reticular @xmath8-@xmath9-@xmath10-@xmath67-induced from @xmath103 .
+ we say that @xmath103 is _ a reticular @xmath8-@xmath9-@xmath10-universal unfolding of _
@xmath67 if @xmath146 is minimal in reticular @xmath8-@xmath9-@xmath10-versal unfoldings of @xmath67 .
+ we say that @xmath103 is _
reticular @xmath8-@xmath9-@xmath10-infinitesimally versal _
if @xmath147 we say that @xmath103 is _
reticular @xmath8-@xmath9-@xmath10-infinitesimally stable _ if @xmath148 we say that @xmath103 is
_ reticular @xmath8-@xmath9-@xmath10-homotopically stable _ if for any smooth path - germ @xmath149 with @xmath150 , there exists a smooth path - germs @xmath151 with @xmath152 and @xmath153 has the form @xmath154 such that each @xmath155 is a reticular @xmath8-@xmath126-@xmath73-isomorphism and @xmath156 for @xmath157 and @xmath158 . let @xmath135 be a neighbourhood of @xmath12 in @xmath136 and let @xmath159 be a smooth map and @xmath32 be a non - negative integer .
we choose a neighbourhood @xmath160 of @xmath12 in @xmath161 such that @xmath162 for any @xmath163 and @xmath164 .
we define the smooth map germ @xmath165 as the follow : for @xmath163 we set @xmath166 by the @xmath32-jet of the map germ @xmath167 at @xmath12 , where @xmath168 is given by @xmath169 + let @xmath99 be an unfolding of @xmath102 .
let @xmath32 be a non - negative integer and @xmath170 .
we say that @xmath103 is _
reticular @xmath8-@xmath9-@xmath10-@xmath32-transversal _ if @xmath171 at @xmath12 is transversal to @xmath172 for a representative @xmath173 of @xmath103 .
* lemma 3.4 ) ) let @xmath99 be an unfolding of @xmath102 . then @xmath103 is reticular @xmath8-@xmath9-@xmath10-@xmath32-transversal unfolding of @xmath67 if and only if @xmath174 [ univunf : tth](uniqueness of universal unfoldings)(cf . , ( * ? ? ?
* theorem 3.13 ) ) let @xmath175 @xmath176 be unfoldings of @xmath177 . if @xmath103 and @xmath104 are reticular @xmath8-@xmath9-@xmath10-universal , then @xmath103 and @xmath104 are reticular @xmath8-@xmath126-@xmath73-@xmath67-equivalent .
[ mthft : th](cf . ,
* theorem 3.14 ) ) let @xmath99 be an unfolding of @xmath102 and let @xmath67 is an unfolding of @xmath74 .
then following are equivalent .
+ ( 1 ) there exists a non - negative number @xmath32 such that @xmath69 is reticular @xmath60-@xmath32-determined and @xmath103 is reticular @xmath8-@xmath9-@xmath10-@xmath178-transversal for @xmath179 , + ( 2 ) @xmath103 is reticular @xmath8-@xmath9-@xmath10-stable , + ( 3 ) @xmath103 is reticular @xmath8-@xmath9-@xmath10-versal , + ( 4 ) @xmath103 is reticular @xmath8-@xmath9-@xmath10-infinitesimally versal , + ( 5 ) @xmath103 is reticular @xmath8-@xmath9-@xmath10-infinitesimally stable , + ( 6 ) @xmath103 is reticular @xmath8-@xmath9-@xmath10-homotopically stable .
we review the theory of reticular legendrian unfoldings .
let @xmath180 be the @xmath181-jet bundle of functions in @xmath182-variables which may be considered as @xmath183 with a natural coordinate system @xmath184 @xmath185 , where @xmath186 is a coordinate system of @xmath187 .
we equip the contact structure on @xmath180 defined by the canonical @xmath181-form @xmath188 .
we have a natural projection @xmath189 by @xmath190 .
we remark that : by the identification of the coordinate system @xmath191 and @xmath192 on a manifold , the dimension of a manifold on which wavefronts propagate is higher @xmath181 than @xmath182 .
we also consider the @xmath181-jet bundle @xmath193 and the canonical @xmath181-form @xmath194 on that space .
let @xmath195 be the canonical coordinate system on @xmath196 and @xmath197 be the corresponding coordinate system on @xmath193 .
then the canonical @xmath181-form @xmath194 is given by @xmath198 we define the natural projection @xmath199 by @xmath200 .
+ let @xmath201 for @xmath202 and let @xmath203 be a representative as a germ of the union of @xmath204 for all @xmath205 .
we recall that @xmath204 is a normalisation of the legendrian submanifold consists of the particles generated by the _ @xmath206-corner _ of hypersurface germ @xmath207 with an @xmath0-corner to conormal direction , where @xmath206-corner is defined by @xmath208 and @xmath209 is a local coordinate system of a manifold in which wavefronts propagate .
let @xmath210 @xmath211 be a contact diffeomorphism germ .
we say that @xmath212 is a _
@xmath126-contact diffeomorphism germ _
if @xmath212 has the form : @xmath213 we say a map germ @xmath214 _ a reticular legendrian unfolding _ if there exists a @xmath126-contact diffeomorphism germ @xmath212 such that @xmath215 is the restriction of @xmath212 to @xmath216 .
we call @xmath217 _ the unfolded contact regular @xmath0-cubic configuration of _ @xmath215 . in order to study perestroikas of wavefronts of reticular legendrian unfoldings , we introduce the following equivalence relation .
let @xmath218 be reticular legendrian unfoldings .
we say that @xmath219 and @xmath220 are _
@xmath126-legendrian equivalent _ if there exist a contact diffeomorphism germ @xmath221 of the form @xmath222 and _ a reticular @xmath0-diffeomorphism _
@xmath223 on @xmath224 such that @xmath225 , where a reticular @xmath0-diffeomorphism is defined by a contact diffeomorphism @xmath223 on @xmath224 such that @xmath226 depends only on @xmath8 and @xmath223 preserves @xmath204 for all @xmath205 .
+ we can construct generating families of reticular legendrian unfoldings .
a function germ @xmath227 is called _ @xmath126-@xmath212-non - degenerate _ if @xmath228 and @xmath229 are independent on @xmath230 .
let @xmath215 be a reticular legendrian unfolding .
a @xmath126-@xmath212-non - degenerate function germ @xmath231 @xmath232 is called _ a generating family _ of @xmath215 if @xmath103 is a generating family of the unfolded contact regular @xmath0-cubic configuration @xmath233 , that is @xmath234 for all @xmath205 . + by ( * ? ? ?
* theorem 0.8.5 ) we have that : [ ucgf : th ] ( 1 ) for any reticular legendrian unfolding @xmath235 @xmath236 , there exists a function germ @xmath227 which is a generating family of @xmath215 .
+ ( 2 ) for any @xmath126-@xmath212-non - degenerate function germ @xmath227 there exists a reticular legendrian unfolding @xmath237 @xmath238 of which @xmath103 is a generating family .
+ ( 3 ) two reticular legendrian unfolding are @xmath126-legendrian equivalent if and only if their generating families are stably reticular @xmath8-@xmath126-@xmath73-equivalent .
let @xmath239 be reticular legendrian unfoldings with @xmath240 where @xmath241 are distinct .
then we call @xmath242 _ a multi - reticular legendrian unfolding_. let @xmath243 and @xmath244 be multi - reticular legendrian unfoldings .
we say that they are reticular @xmath245-legendrian equivalent if there exist contact diffeomorphism germs @xmath246 of the form @xmath247 and reticular @xmath248-diffeomorphisms @xmath249 on @xmath250 such that @xmath251 for @xmath64 .
let @xmath243 be a multi - reticular legendrian unfoldings and @xmath252 be generating families of @xmath253 .
we call @xmath254 _ a multi - generating family _ of @xmath243 .
+ we say that a map germs @xmath255 is _ @xmath245-@xmath212-non - degenerate _ if all @xmath256 are @xmath126-@xmath212-non - degenerate and @xmath257 @xmath258 are distinct .
+ by theorem 0.8.5 in @xcite we have that : @xmath259 for any multi - reticular legendrian unfolding @xmath215 , there exists a multi - generating family of @xmath215 , + @xmath260 for any @xmath245-@xmath212-non - degenerate map germs @xmath261 , there exists a multi - reticular legendrian unfolding of which @xmath103 is a multi - generating family , + @xmath262 let @xmath254 and @xmath263 be multi - generating families of multi - reticular legendrian unfoldings @xmath243 and @xmath264 @xmath265 respectively
. then @xmath243 and @xmath244 are @xmath245-legendrian equivalent if and only if @xmath103 and @xmath266 are stably reticular @xmath8-@xmath9-@xmath10-equivalent .
we define stabilities of multi - reticular legendrian unfoldings .
let @xmath242 be a multi - reticular legendrian unfolding .
+ * stability * : we say that @xmath215 is _ stable _ if the following condition holds : let @xmath267 be @xmath126-contact embedding germs such that @xmath268 and @xmath269 be a representative of @xmath270 . then there exist open neighbourhoods @xmath271 of @xmath272 in @xmath140-topology for @xmath64 such that for any @xmath273 , there exist points @xmath274 such that the multi - reticular legendrian unfolding @xmath275 and @xmath215 are @xmath245-legendrian equivalent , where reticular legendrian unfoldings @xmath276 are chosen that the reticular legendrian unfolding @xmath277 and @xmath276 are @xmath126-legendrian equivalent .
we remark that the definition of stability is not depend on choices of @xmath276 . + * homotopical stability * : we say that @xmath215 is _ homotopically stable _
if for any reticular @xmath126-contact deformations @xmath278 of @xmath253 , there exist one - parameter families of @xmath126-legendrian equivalences @xmath279 on @xmath280 with @xmath281 of the form @xmath282 and a one - parameter deformation of reticular @xmath248-diffeomorphisms @xmath283 such that @xmath284 for @xmath8 around @xmath12 and @xmath64 . +
* infinitesimal stability * : we say that @xmath215 is _ infinitesimally stable _ if for any extension @xmath285 of @xmath253 and any infinitesimal @xmath126-contact transformation @xmath286 of @xmath285 , there exist infinitesimal reticular @xmath248-diffeomorphisms @xmath287 and infinitesimal @xmath126-legendrian equivalences @xmath288 at @xmath289 of the form @xmath290 such that @xmath291 for @xmath64
. we denote @xmath292 ( @xmath146-products ) and denote other notations analogously .
we call an element @xmath293 _ a @xmath245-contact diffeomorphism germ _ if @xmath294 and @xmath295 are distinct .
[ mp_contra : th ] let @xmath296 @xmath297 are submanifolds of @xmath298 . then the set @xmath299 is a residual set in @xmath300 this is directly proved by @xmath126-contact transversality theorem ( ( * ? ? ?
* theorem 0.4.5 ) ) .
+ we denote the ring @xmath301 on the coordinates @xmath302 by @xmath303 and denote other notations analogously .
[ staleg : th ] let @xmath242 be a multi - reticular legendrian unfolding with a multi - generating family @xmath254 .
let @xmath304 be an extension of @xmath215
. then the following are equivalent .
+ ( u ) @xmath103 is a reticular @xmath8-@xmath9-@xmath10-stable unfolding of @xmath305 .
+ ( hs ) @xmath215 is homotopically stable . + ( is ) @xmath215 is infinitesimally stable .
+ ( a ) @xmath306 , where @xmath307 , @xmath308 , and @xmath309 be the @xmath310-submodule of @xmath311 such that @xmath312 @xmath313 for @xmath314 .
we remark that sufficiently near multi - reticular legendrian unfoldings of stable one are stable by the condition ( a ) .
in order to give a generic classification of multi - reticular legendrian unfoldings , we reduce our investigation to a finitely dimensional jet space of @xmath126-contact diffeomorphism germs .
[ l , l+1detleg]let @xmath242 be a multi - reticular legendrian unfolding .
we say that @xmath215 is @xmath32-determined if the following condition holds : for any extension @xmath315 of @xmath253 , the multi - reticular legendrian unfolding @xmath316 and @xmath215 are @xmath245-legendrian equivalent for all @xmath317 satisfying that @xmath318 for @xmath64 . as lemma [ sta : lm ]
, we may consider the following other definition of finitely determinacy of reticular legendrian maps : + ( 1 ) the definition given by replacing @xmath319 to @xmath320 @xmath321 .
+ ( 2 ) the definition given by replacing @xmath319 to @xmath322 .
+ ( 3 ) the definition given by replacing @xmath319 to @xmath323 @xmath321 .
+ then the following holds by proposition 0.5.2 in @xcite : let @xmath215 be a multi - reticular legendrian unfolding .
then + ( a ) if @xmath215 is @xmath32-determined of the original definition , then @xmath215 is @xmath32-determined of the definition ( 1 ) .
+ ( b ) if @xmath215 is @xmath32-determined of the definition ( 1 ) , then @xmath215 is @xmath32-determined of the definition ( 3 ) .
+ ( c ) if @xmath215 is @xmath324-determined of the definition ( 3 ) , then @xmath215 is @xmath32-determined of the definition ( 2 ) .
+ ( d ) if @xmath215 is @xmath32-determined of the definition ( 2 ) , then @xmath215 is @xmath32-determined of the original definition .
let @xmath326 be the set of multi-@xmath32-jets of map germs from @xmath327 to @xmath193 and @xmath328 be the lie group in @xmath326 consists of multi-@xmath32-jets of @xmath126-contact embedding germs .
let @xmath329 be the lie group consists of multi-@xmath32-jets of diffeomorphism germs on @xmath327 .
we consider the lie subgroup @xmath330 of @xmath331 consists of multi-@xmath32-jets of reticular @xmath248-diffeomorphisms on the source space and @xmath32-jets of @xmath126-legendrian equivalences of @xmath332 at @xmath12 : @xmath333 the group @xmath330 acts on @xmath334 and @xmath328 is invariant under this action .
let @xmath335 be a @xmath245-contact diffeomorphism germ from @xmath327 to @xmath193 and set @xmath336 , @xmath337 , @xmath242 , @xmath338 .
we denote the orbit @xmath339 by @xmath340 $ ] . then @xmath341=\ { j^lc'(0)\in tc^l(n , m)\ | \ { \cal l } \mbox { and } ( c'_1|_{{\mathbb l}^{r_1}},\ldots , c'_m|_{{\mathbb l}^{r_m}})\hspace{1cm}\\ \mbox { are $ { \cal p}_{(m)}$-legendrian equivalent } \}.
\end{aligned}\ ] ] we denote by @xmath342 the vector space consists of infinitesimal @xmath245-contact transformation germs of @xmath212 and denote by @xmath343 the subspace of @xmath342 consists of germs which vanish on @xmath12 .
we denote by @xmath344 by the vector space consists of infinitesimal @xmath126-legendrian equivalences on @xmath332 and denote by @xmath345 by the subspace of @xmath344 consists of germs which vanish at @xmath12 .
we denote by @xmath346 the vector space consists of infinitesimal reticular @xmath0-diffeomorphisms on @xmath327 which vanishes at @xmath12 and set @xmath347 .
we have that by lemma [ infsta : t - leglem ] : @xmath348 @xmath349 @xmath350 [ stabletrans_tleg2:th ] let @xmath242 be a reticular legendrian unfolding .
let @xmath361 be an extension of @xmath253 and @xmath362 .
we set @xmath335 .
then the followings are equivalent : + ( s ) @xmath215 is stable .
+ ( t ) @xmath363 is transversal to @xmath364 $ ] .
+ ( a ) @xmath365 , where @xmath307 and @xmath366 @xmath367 , + ( a ) @xmath306 , + ( is ) @xmath215 is infinitesimally stable , + ( hs ) @xmath215 is homotopically stable , + ( u ) a multi - generating family @xmath103 of @xmath215 is reticular @xmath8-@xmath9-@xmath10-stable unfolding of @xmath305 . let @xmath242 be a stable multi - reticular legendrian unfolding .
we say that @xmath215 is _ simple _ if there exists a representative @xmath368 of a extension of @xmath215 such that @xmath369 is covered by finite orbits @xmath370,\ldots,[c_m]$ ] for some multi-@xmath126-contact diffeomorphism germs @xmath371 . in order to classify generic multi - reticular legendrian unfoldings , we classify stable multi - unfoldings @xmath372 of @xmath373 and @xmath374 satisfying the condition : the reticular @xmath9-@xmath375-codimension of @xmath376 for any @xmath164 . +
let a stable multi - unfoldings @xmath372 of @xmath373 and @xmath374 satisfying the condition be given .
then each @xmath256 is a reticular @xmath8-@xmath126-@xmath73-stable unfolding of @xmath377 , there exist monomials @xmath378 such that they consist a basis of @xmath379 and @xmath380 has the maximal degree , and @xmath381 .
then we have that @xmath382 .
since @xmath383 , we have that all @xmath71 are simple singularities .
therefore @xmath69 is stably reticular @xmath60-equivalent to one of the multi - germs in the following list : + @xmath384 ; + @xmath385 , + @xmath386 , + @xmath387 , + @xmath388 ; + @xmath389 , + @xmath390 , + @xmath391 , + @xmath392 .
+ we construct a reticular @xmath9-@xmath10-versal unfolding for each multi - germ by the usual method .
then the corresponding list is as follows : + @xmath393 + @xmath394 .
+ in the case that the reticular @xmath9-@xmath10-codimension of @xmath67 @xmath395 , the @xmath103 is stably reticular @xmath8-@xmath9-@xmath10-equivalent to @xmath396 , where @xmath397 for @xmath398 , @xmath399 , and @xmath400 @xmath401 . in the case that the reticular @xmath9-@xmath10-codimension of @xmath67 @xmath402 , @xmath103 is stably reticular @xmath8-@xmath9-@xmath10-equivalent to @xmath403 , where + @xmath259 @xmath404 @xmath405 , + @xmath260 @xmath406 for @xmath407 , + @xmath262 @xmath408 , + @xmath409 @xmath410 @xmath411 , + @xmath412 @xmath413 for @xmath414 , @xmath415 , and @xmath416 is non - degenerate .
[ genericclass ] let @xmath423 or @xmath181 for @xmath64 and @xmath424 .
let @xmath135 be a neighborhood of @xmath12 in @xmath193 .
then there exists a residual set @xmath425 such that for any @xmath426 and @xmath427 , the multi - reticular legendrian unfolding @xmath428 is stable and have a multi - generating family which is stably reticular @xmath8-@xmath9-@xmath10-equivalent for one of the types in the above list . _
proof_. let @xmath429 be all simple singularities with @xmath430-@xmath73-codim @xmath431 , that is each @xmath432 is one of simple singularities @xmath433 .
for @xmath434 let @xmath435 be @xmath245-@xmath212-non - degenerate map germ which is unfolding of @xmath436 .
we choose an extension @xmath437 and @xmath438 of multi - reticular legendrian unfoldings with multi - generating families @xmath439 and @xmath440 respectively .
let @xmath441 .
we define that @xmath442 \mbox { for all } i=1,\ldots , s \mbox { and } j=0,1\}.\end{aligned}\ ] ] let @xmath443 be all simple singularities with @xmath444-@xmath73-codim @xmath445 .
we choose @xmath446 analogous way .
then we define that @xmath447 \mbox { for all } i=1\cdots , s ' \}.\end{aligned}\ ] ] then @xmath448 and @xmath449 are residual sets .
we set @xmath450>2n+4\}.\ ] ] then @xmath451 is an algebraic set in @xmath452 .
therefore we can define that @xmath453 then @xmath451 has codimension @xmath454 because all @xmath245-contact diffeomorphism germ with @xmath455 adjoin to the above list which are simple . therefore @xmath456 then the set @xmath457 has the required condition.q.e.d
. 99 v.i.arnold,s.m .
gusein - zade & a.n .
varchenko , _ singularities of differential maps , _ vol.i , birkh user , basel,1986 . s.izumiya and g.t.kossioris , _ semi - local classification of geometric singularities for hamilton - jacobi equations _ , j. of differential equations . , * 118*(1995 ) , 166 - 193 s.izumiya , _ perestroikas of optical wave fronts and graphlike legendrian unfoldings _ , j. differential geom . , *
38*(1993 ) , 485 - 500 th : brcker , _ differentiable germs and catastrophes _ , london math .
lecture note ser .
, 17 cambridge univ . press , 1975 .
k.jnich , _ caustics and catastrophes _ , math . ann . , 209(1974 ) , p.161 - 180 .
t.tsukada , _ reticular lagrangian , legendrian singularities and their applications _ , phd thesis , hokkaido university , 1999 .
t.tsukada , _ reticular lagrangian singularities _ , the asian j. of math . ,
vol 1 , * 3*(1997 ) , 572 - 622 .
t.tsukada , _ reticular legendrian singularities _ , the asian j. of math . , *
1*(2001 ) , 109 - 127 .
t.tsukada , _ genericity of caustics and wavefronts on an @xmath0-corner _ , in preparation .
t.tsukada , _ a generic classification of function germs with respect to the reticular @xmath8-@xmath126-@xmath73-equivalence _ , hokkaido math .
j. , 38 , 177 - 203 , 2009 .
t.tsukada , _ bifurcations of wavefronts on an @xmath0-corner _ , in preparation .
g.wassermann , _ stability of unfolding _ , lecture note in mathematics , * 393*. g.wassermann , _ stability of unfolding in space and time _
, acta math . , * 135*(1975 ) , 57 - 128 .
v.m.zakalyukin , _ reconstruction of fronts and caustics depending on a parameter and versality of mappings _ , j. of soviet math . , * 27*(1984 ) , 2713 - 2735 | we give a classification of generic bifurcations of intersections of wavefronts generated by different points of a hypersurface with or without boundaries .
[ section ] [ tth]definition [ tth]lemma [ tth]proposition + takaharu tsukada + college of humanities & sciences , department of mathematics , + nihon university |
there is uncertainty in the bulk velocity of the compact jet .
the doppler enhancement of the intrinsic radiation is a critical parameter in the determination of the jet power , @xmath3 .
the doppler factor , @xmath4 , is given in terms of @xmath5 , the lorentz factor of the outflow ; @xmath6 , the three velocity of the outflow and the angle of propagation to the line of sight , @xmath7 ; @xmath8 $ ] @xcite . for a jet , the apparent flux density ,
@xmath9 is related to the intrinsic value by @xmath10 , where @xmath11 is the spectral index ( the convention adopted in this paper is @xmath12 ) .
assuming an intrinsically symmetric bipolar jet , the elevated flux of the approaching jet has been used to estimate @xmath4 . using a value of @xmath13 that was deduced from discrete relativistic ejecta at other epochs
( see @xcite ) , estimates of @xmath14 for the compact jet at various frequencies and epochs of vlbi observations have been obtained @xcite . for the epoch
april 02 , 2003 , @xcite found disagreement between the jet velocity inferred from the flux asymmetries of the 15 ghz and 8.3 ghz vlba images .
they reconciled this by looking at the clean components , concluding that a value of @xmath15 was compatible with the data .
thus , @xmath16 is the baseline value in the following calculations .
the discussion above is based on direct application of the data to relativistic beaming as was done for discrete ejecta in major flares @xcite . in @xcite ,
the difference in the directly inferred speeds of the compact jet and the major flares was disconcerting .
thus , they conjectured on other explanations including undetected relativistic motion . thus , before the @xcite estimated value was adopted in the jet models ,
an open mind was kept in regards to the value of @xmath6 .
preliminarily , a relativistic jet was considered as suggested in @xcite .
assuming @xmath17 and @xmath18 as was deduced from the kinematics of the discrete ejections in @xcite , it is found that the flux of blackbody radiation from the accretion disk is highly redshifted in the frame of reference of the jet plasma , so synchrotron self compton emission ( ssc ) dominates the external compton emission ( eic ) and ssc is the source of the power law x - rays .
many models were constructed , but because the x - ray luminosity ( of the unresolved base of the bipolar jet ) observed at earth is redshifted by a factor of @xmath19 , the intrinsic x - ray luminosity @xmath20 @xcite .
therefore the jet power , @xmath21 .
this seems unreasonably large .
consequently , relativistic bulk motion is not considered in the following .
these calculations indicate that a reasonable model should have at most mildly relativistic velocities that are consistent with the observed jet asymmetry ( i.e. , @xmath22 ) .
eic is therefore the source of the power law x - ray hard tail ( not ssc ) since the redshifted thermal x - ray flux is much larger than the local synchrotron flux within the base of the jet .
this result is considered in the context of other estimates of the jet velocity in the literature in the discussion section .
in this section , the parameters in the jet model are defined .
one should review @xcite for further discussion .
the jet is parameterized in cylindrical coordinates , @xmath23 .
the jets models are axisymmetric and have no transverse gradients ( this assumption is modified below ) .
the jet shape and axial gradients are parameterized by , @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 , @xmath5 , @xmath30 and @xmath31 through the the expressions , @xmath32 in equation ( 1 ) , the electron energy is @xmath33 and the particle number density is @xmath34 .
the magnetic field strength is @xmath35 .
the jet begins and ends at @xmath24 and @xmath25 , respectively .
the models typically have two regions .
the base of the jet is called the paraboloid , which is the stronger inverse compton source , and it is characterized in @xcite by @xmath36 . the outer part of the jet is called the conical region typically with @xmath37 .
this is generalized to other values of @xmath27 for the outer jet in the following .
the outer jet is the primary source of the radio emission .
tables 1 and 2 list the jet model parameters and physical quantities , respectively , that characterize a jet solution . the dimensional input parameters and power law indices in table 1
define the model in terms of equation 1 .
the chosen value is either fitted to the data or set by the dimensions of the physical system as noted in the table .
all the quantities in table 2 are physical characteristics of the jet model .
it is noted in the table that the values are either an input parameter fitted to the data or a derived physical quantity that is computed from the model .
the jet plasma is protonic .
notice that tables 1 and 2 list four shells and a core ( shell 0 ) instead of one jet .
the reason for this necessity was the surface brightness distribution in the vlba images is very extended . a jet with no transverse gradients that accurately models the sed always produces a surface brightness distribution that decays too rapidly with @xmath38 to represent the vlba maps ( see section 4 ) .
figure 1 shows the relative scale of the nested shells near the base . a continuous transverse gradient would provide a smoother solution
however , the simple crude model of 5 uniform ( there are no transverse gradients within each zone ) zones is shown to be a reasonably accurate representation of the spectrum and the brightness profile in the following .
each shell is parameterized in the tables .
columns 2 , 3 , 4 , 10 and 11 of table 1 define the parameters for the base of the jet in terms of equation ( 1 ) for each shell .
similarly , columns 5 - 9 are the outer jet parameters .
the initial data surface from which the jet solution begins is formally defined for each shell `` i '' as the set of points @xmath39 such that @xmath40 , where @xmath41 is the minimum axial coordinate in shell `` i '' and @xmath42 for @xmath43 .
one important modification to the procedure of @xcite is that the inner shells must transfer radiation not only through the optical depth provided by its own synchrotron self absorption , but also that of the surrounding shells .
this affects the turnover frequency between the optically thick region of the spectrum and the optically thin region of the spectrum through equation ( 11 ) of @xcite .
because the opacity dies off with cylindrical radius , this is a second order correction to the spectrum from each shell , but is retained to improve the accuracy of the calculations . in the following models , @xmath44 is assumed as was determined to be consistent with the kinematics of discrete relativistic ejections in @xcite .
the flux densities are converted to luminosity using a distance of 12.5 kpc to the source as estimated in @xcite .
the baseline fit to the broadband spectral energy distribution ( sed ) is in figure 2 with the x - ray portion magnified in the right hand frame .
these plots are the analog of figure 5 from @xcite .
the x - ray data were calibrated and reduced with current software suite , heasoft v 6.10 and osa v 9.0 , that are improvements from @xcite .
the new results were generously provided by jerome rodriguez .
the fit in the x - ray region is not that good around @xmath45 hz .
there is increased curvature between @xmath46 hz and @xmath45 hz that is more evident in the new reduction of the data than in @xcite .
there is an overlap of the data in the hexte and the integral spectrum in this region as discussed in @xcite .
even though the two data sets do not agree in this area ( the integral spectrum is flatter with higher flux levels , see the related discussion in @xcite ) , this does not appear to be the origin of the spectral curvature , but coincidental .
this region is not well represented by the eic powerlaw from the jet core ( shell 0 ) in blue and is actually exacerbated by the spectral hardening that seems to be real at frequencies above @xmath45 hz . in the model , the hard excess is fit by the red , total luminosity curve which includes a second eic contribution from the surrounding shells . in this decomposition
, the spectral curvature at frequencies just below @xmath45 hz is consistent with spectral ageing from compton losses of the higher energy electrons in the core jet which have the shortest radiative time scale , @xmath47 @xcite .
this suggests that a more elaborate jet model is indicated for the core . in @xcite , the poor powerlaw fit near @xmath45 hz
is improved by including an inverse compton coronal component with a thermal cutoff .
the extra parameters introduced by a corona with cutoff is beyond the scope of this paper , which is an exploration of how well a simple jet model can explain the data .
figure 3 is the surface brightness distribution based on the modeled jet emissivity as it would appear with the resolution provided by the synthetic beamwidth of the vlba as modified by the wavelength dependent interstellar scattering ( @xmath48 mas ) that was determined in @xcite .
the plasma in the outer shells produce the low frequency emission from innermost part of the jet , `` the radio core '' ( or peak flux per beamwidth ) and the observed base of the radio jet ( the partially resolved emission located within one beammwidth of the core , 33.6 au at 8.3 ghz and 17.0 au at 15 ghz ) that is depicted by the brightness distribution in figure 3 .
shell 1 produces the farthest reaches of the jet .
the lowest contour is at 15 times the rms noise of the radio images , thus the vertical error bars are very small on all the data points .
the dominant source of error is the sparse u - v coverage ( the missing 8.3 ghz flux compared to the vla measurement noted in the figure caption ) .
there is really no way to attach these errors to any of the individual contours .
all that one can say is that the relatively long baselines of the vlba favors missed flux on the least compact scales , the outer - most contours . [
cols="^,^,^,^,^,^,^,^,^,^,^ " , ] the innermost shell labeled `` 0 '' ( the core ) is initiated in the innermost part of the accretion disk . in the model
, it resides in the ergospheric portion of the disk inside of 1.7 m ( @xmath49 cm in geometrized units for the 14 @xmath50 black hole mass inferred by @xcite ) .
thus , one requires a rapidly spinning black hole and a black hole spin of @xmath51 is necessary to provide sufficient surface area in the disk .
if a slowly spinning black hole is used in the model , the volume of the base of the jet increases dramatically and so does @xmath3 ( which is highly dependent on the total amount of energized particles and magnetic field - large jet volumes create large total plasma energy in the models ) .
the magnetic field is turbulent and it is assumed to be advected with the jet as in @xcite .
the jet begins near equipartition as noted in table 2 . from equation 1
, the value of @xmath52 means that magnetic energy is dissipated rapidly in the plasma near the base of the jet .
the lost energy is the x - ray emission forming the high energy power law .
the core is the most powerful portion of the jet and most of the energy is lost to x - rays by inverse compton cooling .
the x - ray spectrum is computed per the methods of @xcite .
the x - ray power law spectral index , @xmath53 , fixes @xmath54 .
this is a very steep value that is consistent with the assumption of intense compton cooling .
the disk emission is fit by a simple blackbody ( instead of a parameterized accretion disk model ) with @xmath55 k and a surface area on each disk face of @xmath56 ( the surface area of a thin keplerian disk for @xmath57 and @xmath58 ) .
the spectral fit has @xmath59 ergs / sec in thermal disk luminosity and @xmath60 ergs / sec in jet luminosity ( note the total matches that inferred by @xcite when the flux overestimate from rxte / pca of 16% that they note is taken into account ) .
the large euv peak in the synchrotron spectrum in figure 2 is from the same electrons that produce the eic .
the outer portion of the core jet at large z is inertially dominated and is a weak synchrotron source .
shells 1 to 4 are poynting flux dominated jets with a field line angular velocity , @xmath61 , equal to the relativistic keplerian velocity , @xmath62 , at the foot point @xcite .
the field is organized and satisfies the perfect mhd condition similar to the electrodynamic models of @xcite , @xmath63 the jets have an artificial simplification that the poloidal bulk velocity is constant , @xmath64 . from equations ( 1 ) and ( 2 ) , at large @xmath38 , the toroidal magnetic field , @xmath65 , is much larger than the poloidal magnetic field , @xmath66 , and the particle angular velocity , @xmath67 , is small compared to @xmath61 ( bounded velocity and angular momentum ) .
perfect mhd requires that the total poloidal magnetic flux is conserved in each shell , @xmath68 where @xmath69 is the cross - sectional area in the shell .
equations ( 1 ) - ( 3 ) combine to give a relationship between @xmath66 at the base of the jet and @xmath65 at the start of the outer jet , @xmath70 this relationship allows one to compute ( as in table 2 ) @xmath71 and the poloidal poynting flux , @xmath72 , in both regions of the jet , where @xmath73 the base of the shell 1 jet produces the bulk of the synchrotron ir emission .
the jet emission in @xmath74 band was estimated to be about 40% of the total luminosity ( the bulk being from the donor star ) in @xcite consistent with the fit in figure 2 . the value of @xmath75 for the shells is chosen to reproduce the noticeable hardening of the x - ray spectrum above @xmath45 hz seen in the integral data in figure 2 and discussed above .
more than one half of this excess over the steep core powerlaw arises from eic at the base of the shell 1 jet .
the other shells provide the remainder of the hard eic excess and the inner part of the brightness profile of the jet in figure 3 .
shell 4 is very energetic because the volume is huge - it has a very wide base .
shells 2 and 3 are required to interpolate the brightness distribution in figure 3 between the `` radio core '' and base of the observed radio jet produced by the plasma in shell 4 and the distant regions of the brightness distribution that is emitted from plasma in shell 1 . in summary , if one assumes that the hard x - ray power law is produced in the jet ( as in the fiducial model described in the last two sections ) then the results in table 2 imply that the jet power is @xmath76 ergs / sec .
assuming @xmath58 , the mass flux in the jet is @xmath77 g / s .
it is not claimed that this fiducial model is the unique jet solution that fits the data .
this raises the question of how the model depends on the parametric assumptions .
although a formal parametric sensitivity analysis is beyond the scope of this work , many trial models were produced and evaluated in the process of arriving at the model presented in the previous sections . here
are some of the observations that were made . * * number of shells * the integrated kinematic quantities such as the magnetic , energy and mass fluxes are not significantly affected by the number of shells . the more shells , the smoother the brightness distribution and the better the fit is in figure 3 . for example , if 4 or 6 shells were used instead of 5 , the sed can be reproduced
. however , the fit in figure 3 , would be considerably worse for 4 shells and somewhat better for 6 shells .
the luminosity in each shell decays on scales that are comparable to the distance , @xmath38 , that the peak of the luminosity is from the accretion disk .
for example , the 15 ghz luminosity is produced at smaller @xmath38 than the 8.3 ghz luminosity in each shell .
thus , the surface brightness dies off with a shorter scale length at 15 ghz than at 8.3 ghz .
compare the right hand frame of figure 3 , to the left hand frame - the 15 ghz fit is worse in the sense that there are large curved sections between actual data points and a pronounced inflection point neither of which exist within the sparsely sampled data ( there are only 6 contour levels in 15 ghz image in @xcite ) .
this is manifested as the almost exponential decay near 10 au , followed by a sharp peak at @xmath78 25 au .
the more shells , the less pronounced the alternating regions of exponential dips and sharp peaks become . with 4 shells , the 8.3 ghz plot would have similar large deviations from the brightness profile as are seen in the existing 15 ghz plot in figure 3 . *
* energy flux estimate * given the assumptions in the model ( in order of importance : the hard x - ray luminosity is from the jet , the hard x - ray excess above @xmath45 ghz is real and from the jet , and the ir excess above the stellar contribution is from the jet ) , the model is very close to a minimum energy flux .
this is obvious since the energy flux is only 1.5 times the radiated luminosity and the jet power must exceed the broadband luminosity .
the energy flux grows very rapidly , if the jet is assumed to originate at large cylindrical radius .
the volume of plasma and fields increases rapidly and the energy flux quickly rises above @xmath79 ergs / sec as the cylindrical radii of the base of the shells increase by a factor of @xmath80 . *
* number density * the number of energetic particles is fixed by the location of the base of the jet , the thermal x - ray flux from the accretion disk and the constraints on the observed external compton luminosity . the particle flux and energy flux are concentrated at small radius in the core .
driving most of the energy and mass flux from the inner region of the disk is favored on physical grounds .
it co - locates the most energetic shell and largest mass flux at the largest potential source of power to drive the jet - the most energetic part of the accretion flow ( fastest rotation rate and highest temperature ) . moving the core to the smallest possible cylindrical radius
also moves the other shells inward to smaller radius and larger keplerian velocity , thereby decreasing the total volume and magnetic flux required to drive the flow , by equation ( 5 ) .
this minimizes the energy flux in the outer shells . in this decomposition , the mass flux in the outer shells is constrained by the magnitude of the hard x - ray excess above @xmath45 hz ( eic from shells 1 -4 ) discussed in section 2 .
the large radio fluxes from the outer shells combined with a modest eic x - ray component result in the magnetic dominance for the outer shells depicted in columns 7 and 8 of table 2 .
this analysis does not preclude the possibility of a different transverse distribution of mass flux in a jet that can reproduce the data , however the energy flux is likely to be larger . * * magnetic field strength * the magnetic field strength , @xmath35 is set by the size of the base of the jet , the particle number density and the synchrotron portion of the sed .
the magnetic field must also be of sufficient strength to be the ultimate energy reservoir that supports the bulk of radiation losses as discussed in section 3 . * * electron energy spectrum * the electron spectrum is fixed by the x - ray power law spectrum as mentioned in section 2 .
the only parameter that is not affected by the data is @xmath81 .
since the energy distributions are steep , there are very few high energy electrons .
a value of @xmath82 was chosen arbitrarily .
any value larger than 700 will show negligible differences in the results . * * axial dimensions * the value of @xmath41 is set by the physical dimensions of the accretion / disk black hole system .
the quantities defined at @xmath41 form the initial data for the initial surface of the entire jet . by contrast , @xmath83 is set by the sed fit .
for example , large values of @xmath83 over - produce the high frequency synchrotron emission . by definition @xmath84 .
the solution is insensitive to @xmath85 cm , since the jet emissivity is very low in this region .
the shell decomposition in table 2 is very useful for exploring variants of the jet model .
the most obvious variant is to assume that the disk corona and not the jet is the source of the x - ray powerlaw . by lowering the mass density and magnetic field density significantly ( say , a factor of 100 for the sake of argument ) in the core
, one can simply ignore the energy flux and luminosity contribution to the composite jet . then summing the remaining components in table 2 , @xmath86 and the jet plasma still reproduces the sed ( less the coronal x - rays ) and the radio brightness profile in figures 2 and 3 , respectively .
there is uncertainty in the non - relativistic jet velocity as noted in section 1 .
thus , it desirable to know the sensitivity of the jet power to different values of jet poloidal velocity , @xmath87 .
the shell model lends itself to estimating @xmath3 for different values of @xmath88 .
first decompose @xmath3 into 3 discrete pieces . then note from from equations ( 2 ) and ( 5 ) , that @xmath72 in the outer jet has a simple scaling with @xmath88 , @xmath89 .
since @xmath3 is a sum of radiation losses , the mechanical kinetic energy flux in the outer jet , @xmath90 , and @xmath91 , one can use the results in table 2 to parameterize @xmath3 as a function of @xmath88 when the x - ray power law is from the jet . implementing the doppler enhancement factors from @xcite , @xmath92\nonumber \\ & & \times ( 2.5\times 10^{38 } ) \nonumber \\ & & + \left[\frac{\delta(\beta=-0.4,\ ; \theta = 0^{\circ})}{\delta(-\beta,\ ; \theta = 0^{\circ})}\right]^{4 } ( \gamma- 1)(v^{p}/0.436 c)(8.5 \times 10^{37})\nonumber \\ & & + ( v^{p}/0.4c)(8.3 \times 10^{37 } ) \ , \mathrm{ergs / sec},\end{aligned}\ ] ] where @xmath93 , the doppler factor that was introduced in section 1 , is a two parameter function of @xmath6 and @xmath7 .
in this notation , @xmath94 means to evaluate the doppler factor at @xmath95 and @xmath96 .
the jet poloidal velocity is @xmath97 .
the doppler correction on the first term in equation ( 7 ) , means that larger @xmath4 factors require smaller intrinsic radiation losses to produce the observed x - ray powerlaw sed in figure 2 .
the doppler correction on the second term results from the redshifting of the disk thermal emission in the rest frame of the plasma at the base of the jet .
a slow jet sees a stronger seed field and requires fewer hot particles to reproduce the eic sed and therefore has a smaller bulk kinetic energy flux .
the last term is the electromagnetic energy flux in the outer jet .
as noted in section 1 , this formula will not hold for large @xmath6 since ssc contributions become important .
similarly , if the disk corona is the source of the x - ray powerlaw , one can use the results in table 2 and equation ( 6 ) to parameterize the jet power as a function of jet velocity as @xmath98[8.3 \times 10^{37}]\ , \mathrm{ergs / sec}\;.\end{aligned}\ ] ]
the crude model presented here has demonstrated a few facts about the compact jet in grs 1915 + 105 in the `` high plateau state '' : * it is possible to model the jet in a manner analogous to agn * the jet is powered during a high accretion state with a thermal luminosity of @xmath99 ergs / sec , corresponding to an eddington rate of @xmath78 23% . *
the jet brightness profile requires a highly stratified jet in the transverse direction * relativistic jet bulk velocity is not favored based on the huge implied energy flux , @xmath100 ergs / sec * if * the jet is the source of the x - ray power law .
an interesting consequence of the model was that a rapidly spinning black hole was preferred on energetic grounds .
this was not to tap black hole spin energy , but to use the surface area and high angular velocity of the plasma in the ergosphere to power the jet core .
it is interesting to note that the 3-d , perfect mhd , general relativistic numerical simulations in @xcite find very high efficiency `` funnel wall jets '' emanating from this region of the ergosphere of high spin black holes .
the outflow is somewhat similar to what is described in the core shell in this treatment , near equipartition with turbulent disordered magnetic flux and an outflow velocity of @xmath101 0.3c .
the efficiency is sufficient to power the core if @xmath102 .
they argue that the power source is pressure gradients driven by the disk which seems energetically reasonable in grs 1915 + 105 considering the large thermal luminosity noted above .
even though there is some numerical evidence to support the notion of a near - relativistic jet core ( where most of the energy transport occurs ) , it is not clear if there is any analog of the compact jet in the agn population . in particular , are there near - relativistic outflows that are highly stratified in the transverse direction that carry a mass flux comparable to the accretion flux ?
it is also not obvious if there is any structure that can be construed as an analog of a non - relativistic high luminosity , optically thick , radio jet in agn .
if there is scale invariance in black hole accretion systems then there should be an analog of the compact jet in some agn .
the only obvious way to reconcile this difference is to consider the possibility that an agn radio jet analogy is not appropriate and the analog of the compact jet in agn takes a different guise due to the lower accretion disk temperature .
the agn feature that possesses the majority of the characteristics of the jet model of grs 1915 + 105 are the near - relativistic outflows that are seen in x - ray absorption .
a few radio quiet quasars have been observed to have out - flowing , x - ray absorbing winds with velocities between 0.1 c and 0.6 c @xcite .
the outflow in apm08279 + 5255 is the most energetic with a mass flux similar to the accretion mass flux and a kinetic luminosity comparable to the thermal luminosity of the accretion flow as in grs 1915 + 105 @xcite . the fastest known out - flowing , x - ray absorbing winds in radio loud agn have @xmath103 @xcite . in summary , due to the relatively low jet velocity and the large transverse stratification ,
perhaps the compact jet in black hole binaries is best described as a bipolar near - relativistic wind . in this analogy , it is not obvious why the galactic black hole version of the wind is luminous in the radio band and the agn version is not
. it has been argued in @xcite that there may exist a compact , optically thick jet in low - luminosity agn .
this may or may not be applicable to the study here in that the thermal luminosity was 23% of the eddingtom limit in these observations of grs 1915 + 105 . in any event , it is worthwhile to study jets from low luminosity agn with vlbi to look for evidence of nonrelativistic optically thick winds or jets .
the implication of these jet models to other galactic black holes can only be speculated upon because other sources do not have the wealth of high resolution radio images .
furthermore , grs 1915 + 105 is a very strong x - ray source , so there are many excellent studies of the radio jet/ x - ray connection .
consider , the results of @xcite that the existence of a hard power - law component is tantamount to the existence of a jet .
within the framework of the assumption that inverse compton emission in the jet is the dominant source of the power law component , the jet power can be estimated from the model presented here as approximately 1.5 times the hard x - ray power law luminosity .
this estimate can be applied to other galactic black holes in the hard state .
however , it would be much more accurate to constrain the energy flux derived from the basic jet model on a case by case basis with deep radio images and broadband simultaneous spectroscopy , which are not readily available for other galactic black holes . there are estimates of the compact jet energy flux based on less direct methods for other black hole candidates such as cygnus x-1 .
for example , the emission from a nebular ring 5 pc from cygnus x-1 has been used to estimate the energy flux required to energize the structure .
this in turn is related to the compact jet power on much smaller scales by assuming an invisible flow of energy to large scales .
this leads to estimates of the kinetic power of the compact jet of @xmath104 to @xmath105 ergs / sec @xcite .
this estimate is 0.3 to 1.0 of the total radiative luminosity , which is similar to the jet power of approximately 0.6 times of the total radiative luminosity that was found for the compact jet in grs 1915 + 105 here .
another consequence of the model is that a relativistic outflow velocity in compact jets is not favored based on energetic concerns .
it is interesting to compare this to less direct estimates of the compact jet speed in grs 1915 + 105 and other galactic black hole candidates .
using the very basic jet model of @xcite , it is argued in @xcite that @xmath106 in gx 339 - 4 , which was have ruled out for grs 1915 + 105 in the epoch considered here .
a very general argument was presented in @xcite that was based on total x - ray and radio fluxes from a one zone model ( that happens to be cylindrical in shape ) .
they found a loose set of constraints that favored @xmath106 in galactic x - ray binaries , but was also consistent with @xmath107 .
contrary to these works @xcite used monte carlo simulations to argue that @xmath108 is preferred for galactic x - ray binaries in the low / hard state , marginally consistent with the energetic constraints of the models presented here .
i am indebted to jerome rodriguez for analyzing the x - ray data and discussing the implications with me .
i would also like to thank marc ribo for useful discussions regarding the radio data .
i was lucky to have a referee who offered many suggestions that improved the presentation of the manuscript .
99 blandford , r. d. 1976 , mnras * 176 * 465 blandford , r. and konigl , a. 1979 , apj * 232 * 34 casella , p. et al , 2010 , mnras lett .
* 404 * 21 chartas , g. , brandt , w.n . , gallagher , s. 2003 , apj * 595 * 85 dwahan , v. , mirabel , i.f . , rodriguez , l. 2000 , apj * 543 * 373 fender , r. et al , 1999 , mnras * 304 * 865 fuchs , y. et al , 2003 , a and a * 409 * l35 fuchs , y. et al , 2004 , proc . of the 5th integral workshop ( esa )
gallo , e. , fender , r. , pooley , g. 2003 , mnras * 344 * 60 gallo , e. et al 2005 , nature * 436 * 819 ghisellini , g. , maraschi , l. and treves , a. , 1985 , a & a * 146 * 204 ghisellini , g. and maraschi , l. , 1989 , apj * 340 * 181 ghisellini , g. , maraschi , l. and dondi , l. , 1996 , suppl . series a & a * 120 * 503 gofford , j. , et al 2011 , mnras * 414 * 3307 greiner , j. , cuby , j. , mccaughrean , m. 2001 , nature * 414 * 522 hawley , j. , krolik , k. 2006 , apj * 641 * 103 heinz , s. , merloni , a. 2004 , mnras lett . * 355 * l1 koerding , e. , jester , s. , fender , r. 2006 , mnras * 372 * 1366 lightman , a. , press , w. , price , r. and teukolsky , s. 1975 , _ problem book in relativity and gravitation _ ( princeton university press , princeton ) lind , k. , blandford , r. 1985 , apj * 295 * 358 lovelace , r. v. e. 1976 , nature * 262 * 649 markoff , s. , nowak , m. , wilms , j. 2005 apj * 635 * 1023 mirabel , i.f . , rodriguez , l. 1994 , nature * 371 * 46 murray , n. , chiang , j. , grossman , s. , voit , g. 1995 apj * 451 * 198 ogley , r. < bell burnell , s. fender , r. , pooley , g. , waltman , e .
2000 , mnras * * 317**158 pooley , g. and fender , r. 1997 , mnras * 292 * 925 pounds , k. , king , a. , page , k. , obrien , p. 2003
, mnras * 346 * 1025 reeves , j. , et al 2009 , apj * 701 * 493 ribo , m. , dhawan , v. mirabel , f. 2004 , proc . of the 7th vlbi network symposium , bachiller , r. , colomer , f. , desmurs , j. , de vicenete , p. ( eds ) toledo , spain astro - ph/0412657 rodriguez , j. , et al 2004 , apj * 615 * 416 rodriguez , j. , et al 2008 , apj * 675 * 1436 rodriguez , l. , mirabel , i.f . 1999 , apj * 511 * 398 rushton , a. , spencer , e. , fender , r. and pooley , g. 2010 , a & a * 524 * 29 rushton , a. , spencer , e. , pooley , g. , trushkin , s. 2010 , mnras * 401 * 2611 russell , d. , fender , r. , gallo , e. , kaiser , c. 2007 , mnras * 376 * 1341 saez , chartas , g. , brandt , w.n .
2009 , apj * 697 * 194 sambruna , r. et al 1997 , apj * 474 * 639 tombesi , f. , et al 2010 , apj * 719 * 700 tucker , w. 1975 , _ radiation processes in astrophysics _
( mit press , cambridge ) . | in this article , models are constructed of the compact jet in grs 1915 + 105 during an epoch of optimal data capture . on april 02 , 2003 , the object was observed in the hard x - ray / soft gamma ray band ( integral ) , hard x - ray band ( rxte ) , near ir ( eso / new technology telescope ) and the vlba ( 8.3 ghz and 15 ghz ) . the source was in a so - called `` high plateau state . ''
the large radio flux provides high signal to noise ratios in the radio images .
thus , one can image the jet out to large distances ( @xmath0 cm ) .
this combined with the broadband coverage make this epoch the best suited for modeling the jet . the parametric method developed in the papers @xcite that has been successfully utilized in the realm of extragalactic radio jets
is implemented .
the basic model is one where external inverse compton ( eic ) scattering of accretion disk photons by jet plasma provides the hard x - ray powerlaw . unlike agn jets
, it is found that the radio jet must be highly stratified in the transverse direction in order to produce the observed surface brightness distribution in the radio images .
various jet models are considered .
the jet power is @xmath1 ergs / sec if the hard x - ray powerlaw luminosity is from eic in the jet and @xmath2 ergs / sec if the x - rays are emitted from the accretion disk corona .
these estimates indicate that the jet power can be as high as 60% of the total x - ray luminosity .
[ firstpage ] black hole physics x - rays : binaries accretion , accretion disks galactic black hole accretion systems might serve as laboratories for agn ( active galactic nuclei ) on highly contracted time scales . the black hole candidate , grs 1915 + 105 , routinely ejects powerful radio emitting plasma out to large distances from the central black hole at relativistic speeds and might be the galactic analog of radio loud agn @xcite .
these major flares are relatively rare compared to the so - called compact jet state @xcite .
a compact , optically thick , elongated structure is frequently observed with long baseline interferometry @xcite .
the low frequency emission is significantly less than in a major flare .
it is not clear if there is an analogous structure in the context of agn .
the total power of the jet is also unclear . in this article , the analogy with agn
is pursued by modeling the jet using the standard agn technique of @xcite with the addition of external compton scattering ( eic ) from @xcite .
there is a correlation or more precisely a direct association of the formation of a compact jet with an excess of hard x - ray emission in the form of a power law @xcite .
the primary assumption of the models of the jet presented below is that this x - ray emission is produced in the base of the jet @xcite .
a scenario in which the x - rays are from compton scattering in the accretion disk corona is also explored briefly in this article with the same techniques .
since most of the luminosity is in the x - ray band , complete high energy coverage is mandatory for jet modeling .
furthermore , one needs the near ir photometry in multiple bands in order to constrain the synchrotron spectrum . also , to break the degeneracy in the available models , one requires a faithful representation of the surface brightness distribution of the jet .
it is a much simpler task to fit the jet spectrum if the distribution of emissivity is ignored . to meet this challenge in the presence of intrinsic source variability
, one needs simultaneous full coverage of the x - ray emission to high frequencies , ir photometry and sensitive vlba imaging .
the only simultaneous set of observations capable of fulfilling these requirements was the campaign presented in @xcite .
the strongest compact jet ever imaged with vlba was measured simultaneously in the near ir ( eso / new technology telescope ) and throughout the hard x - ray and soft gamma ray band ( integral and rxte ) . for more details on the observations
see @xcite . |
there exists an ever - growing trend in high ( or `` big '' ) dimensional data processing to design new procedures ( or to simplify existing ones ) using linear dimensionality reduction ( ldr ) methods in order to get faster or memory - efficient algorithms .
provided this reduction does not bring too much distortion between the initial data space and the `` reduced '' domain , as often allowed by the intrinsic `` low - dimensionality '' properties of the input data , many techniques , such as nearest - neighbor search in big databases @xcite , classification @xcite , regression @xcite , filtering @xcite , manifold processing @xcite or compressed sensing @xcite can be developed in this reduced domain with controlled loss of accuracy , as well as stability with respect to data corruption ( _ e.g. _ , noise ) .
most often , those ldr tools rely on defining a random projection matrix ( sometimes called _ sensing _ matrix ) with fewer rows @xmath7 than columns @xmath13 , whose multiplication with data represented as a set of vectors in @xmath14 provides a reduced representation ( or _ sketch _ ) of the latter .
this is the scheme implicitly promoted for instance by the celebrated johnson - lindenstrauss ( jl ) lemma for finite sets of vectors @xmath15 , _
i.e. _ , with @xmath16 @xcite . this cornerstone result and its subsequent developments @xcite showed that , given a resolution @xmath17 , if @xmath18 where @xmath19 is the cardinality of @xmath20 and @xmath21 is a general constant , then a random matrix @xmath22 whose entries are independently and identically distributed ( ) as a centered sub - gaussian distribution with unit variance defines an isometric mapping that preserves pairwise - distances between points in @xmath20 up to a multiplicative distortion @xmath23 .
in other words , @xmath24 defines an @xmath23__-isometry _ _ between @xmath25 and @xmath26 , _
i.e. _ , with high probability , for all @xmath27 , @xmath28 equivalently , one observes that keeping the probability of success constant with respect to the random generation of @xmath24 and inverting the requirement linking @xmath7 and @xmath23 , such an isometry has a distortion @xmath23 decaying as @xmath29 when @xmath7 increases , _
i.e. _ , this distortion vanishes when @xmath30 is large .
notice that variants of this embedding result exist with different `` input / output '' norms ; see , _
e.g. _ , @xcite for a unified treatment over a family of _ interpolation _ norms including @xmath6 and @xmath5 as special cases .
the jl lemma has been later generalized to any subsets @xmath31 , not only finite , whose typical `` dimension '' can be considered as small with respect to @xmath13 ( see , _
e.g. _ , @xcite ) .
in other words , as soon as @xmath32 displays some internal structure that makes it somehow parametrisable with much fewer parameters than @xmath13 , as for the set of sparse or compressible signals , the set of low - rank matrices , signal manifolds , or a set given as a union of low - dimensional subspaces , an @xmath23-isometry like can be defined for all pairs of vectors in @xmath32 .
this is for instance the essence of the restricted isometry property ( rip ) and its link with the jl lemma , where holds with high probability for all @xmath33-sparse vectors provided @xmath34 @xcite .
however , these embeddings have one strong limitation . except in very specific situations , such as for discrete sub - gaussian random matrices @xmath24 ( _ e.g. _ , bernoulli ) and finite sets @xmath32 ,
the set @xmath35 is not finite .
an infinite number of bits is thus required if one needs to store , process or transmit @xmath36 without information loss for any possible @xmath37 . moreover , knowing how many bits are required to represent such projections is also important theoretically for assessing and measuring the level of information contained in the reduced data space or for improving specific data retrieval and processing algorithms .
additionally , if this measure of information can be achieved , nothing prevents us to take @xmath38 , as the sought `` dimensionality reduction '' can be aimed at minimizing the number of bits rather than the dimensionality @xmath7 .
for instance , @xcite defines locality - sensitive hashing ( lsh ) as a procedure to turn data vectors into quantized _ hashes _ that preserve locality , so that close vectors induce , with high probability , close hashes .
however , this method is specifically designed for boosting nearest - neighbor searches over a finite set of vectors and not to define an isometry similar to . as a more practical solution ,
the embedding realized by a random projection @xmath24 is often followed by a scalar quantization procedure , _
e.g. _ , with a uniform scalar quantizer @xmath39 with resolution @xmath40 , applied componentwise on the image of @xmath24 .
a direct impact of this sequence of operations is to induce a new additive distortion in related to @xmath41 , as discussed in @xcite . indeed , assuming @xmath24 respects for all @xmath42 and @xmath43 in a certain subset @xmath31 , given a uniform quantizer @xmath44 of resolution @xmath40 applied componentwise on vectors of @xmath45 we would have @xmath46 for all @xmath47 , which involves @xmath48 for any @xmath49 .
therefore , a simple manipulation of provides @xmath50 in other words , as described in sec .
[ sec : framework ] , the quantized mapping @xmath51 defines now a _
quasi - isometric _ embedding between @xmath52 and @xmath53 . however , while displays a constant additive distortion , several works in this context have observed that such an additive error actually decays as @xmath7 increases .
first , when distances in the reduced space are measured with the @xmath5-norm and when @xmath54 is combined with a _
_ dithering _ _ and eq . ) . ] , a quasi - isometry similar to holds with high probability for all vectors in a _
set @xmath55 @xcite .
the additive distortion reads then @xmath56 for some absolute constant @xmath57 and this error also decays as @xmath58 , as does the multiplicative error @xmath23 .
second , when combined with universal quantization @xcite , _
i.e. _ , with a periodic scalar quantizer @xmath54 , an exponential decay of this distortion as @xmath7 grows can be reached ; for the moment , this has been proved only for sparse signal sets .
finally , recent works related to 1-bit compressed sensing ( cs ) have shown that for a quantization @xmath54 reduced to a sign operator ( _ i.e. _ , @xmath59 ) the angular distance between any pair of vectors of a low - dimensionality set @xmath32 is close to the hamming distance of their mappings up to an additive error decaying as @xmath60 for some @xmath61 .
this is true for random gaussian matrices and for the set of sparse signals @xcite , for any sets with `` low dimensionality '' as measured by their gaussian mean width @xcite ( see below ) and even for sub - gaussian random matrices provided the projected vectors are not `` too sparse '' @xcite , _ i.e. _ , for vectors whose @xmath62-norm is much smaller than their @xmath6-norm .
* contributions : * considering these last observations , the main results of this paper show that : + * quasi - isometric embeddings can be obtained with high probability from scalar ( dithered ) quantization after linear random projection ; for such embeddings both multiplicative and additive distortions co - exist when , as in @xcite , distances between mapped vectors are measured with the @xmath5-norm - distance . ] ; * random sensing matrices for such embeddings are allowed to be generated from symmetric sub - gaussian distributions provided embedded vector differences are not `` too sparse '' ( as in the 1-bit case @xcite ) ; * the results above actually hold with high probability for _ any _ subset @xmath32 of @xmath14 as soon as @xmath7 is large compared to its typical dimension , _
i.e. _ , to its squared gaussian mean width . * with high probability , the biggest distance separating two _ consistent _ vectors in @xmath32 ( _ i.e. _ , characterized by identical quantized mappings ) , that is what we call the _ consistency width _ , decays when @xmath7 increases at a faster rate than what could be predicted by using just the implications of a quasi - isometry .
this extends to any set @xmath32 the works of @xcite , that were valid only for sparse signals ; * for particular _ structured _ sets , _
e.g. _ , the set of ( bounded ) sparse vectors or the set of ( bounded ) low - rank matrices , the minimal values of @xmath7 necessary to specify a quantized embedding or a small consistency width can be strongly reduced compared to those required for a general set ; moreover , we aim at optimizing whenever it is possible the requirements on @xmath7 ( _ e.g. _ , with respect to @xmath23 and @xmath41 ) that guarantee those results .
* methodology : * as an important aspect of our developments , we study the conditions for obtaining quasi - isometric embeddings of any bounded subsets @xmath31 into @xmath63 . following key procedures established in other works @xcite , the typical dimension of these sets
is measured by the _
gaussian mean width _ , _
i.e. _ , @xmath64 with @xmath65 .
this quantity , also known as gaussian complexity , has been recognized as central for instance in characterizing random processes @xcite , shrinkage estimators in signal denoising and high - dimensional statistics @xcite , linear inverse problem solving with convex optimization @xcite or classification efficiency for randomly projected signal sets @xcite .
more specifically , the minimal number of measurements @xmath7 necessary to induce , with high probability , an @xmath66-isometric embedding of any subset @xmath67 into @xmath45 from sub - gaussian random projections is known to be proportional to @xmath68 @xcite . therefore , since @xmath69 for some finite set @xmath32 , we recover the condition defining the johnson - lindenstrauss lemma by imposing latexmath:[$m \gtrsim
\log @xmath33-sparse vectors in an orthonormal basis ( onb ) @xmath71 , @xmath72 , which characterizes the conditions of the restricted isometry property ( rip ) for sub - gaussian random matrices @xcite . the interested reader can find a summary of the main properties of the gaussian mean width in table [ tab : gaussian - mean - width - prop ] , with explicit references to their origin .
this table could be helpful also to keep trace of these properties while reading our proofs . in our developments ,
we sometimes complete the characterization of sets provided by the gaussian mean width with another important measure : the kolmogorov @xmath23-entropy of a set @xmath31 that we denote @xmath73 @xcite .
this is defined as the logarithm of the size of the smallest @xmath23-net of @xmath32 , _
i.e. _ , a set @xmath74 such that any vector of @xmath32 can not be farther than @xmath23 from its closest vector in @xmath75 . by the sudakov inequality
, this entropy is connected to the gaussian mean width as @xmath76 .
however , in specific cases this last inequality is too loose with respect to @xmath23 .
as summarized in @xcite , this is the case of the _ structured sets _
@xmath32 defined hereafter , for which this work will provide separated and tighter results .
[ def : structured - set ] a bounded set @xmath31 with diameter @xmath77 is _ structured _ iff there exists a quantity @xmath78 , independent of @xmath79 , for which we have both [ eq : structured - set - prop ] @xmath80 for any @xmath81 , where @xmath82 is the _ local set _ of @xmath32 of radius @xmath83 .
for instance , if @xmath84 is a subspace of @xmath14 , a union of subspaces ( such as the set @xmath85 of @xmath33-sparse signals in an orthonormal basis or in a redundant dictionary @xmath86 of @xmath14 ) , the set of rank-@xmath87 matrices @xmath88 in @xmath89 , or even the set of group - sparse signals , then @xmath84 is a _ cone _ , _
i.e. _ , @xmath90 for any @xmath91 , and the set @xmath92 is structured for any diameter @xmath93 @xcite . indeed ,
focusing first on , if @xmath84 is one of the sets listed above , @xmath94 is also a cone and @xmath95
. therefore @xmath96 . this last quantity is easily bounded since @xmath97 often shares the same structure than @xmath84 , _
e.g. _ , @xmath98 if @xmath99 , and in fact @xmath100 showing that @xmath78 can be set to @xmath101 in .
second , for , the komogorov entropy of such a set @xmath84 can often be tightly bounded by decomposing it into a union of subspaces or subdomains restricted to @xmath102 , so that a global @xmath23-net of small cardinality could be reached by the union of the @xmath23-nets of all of these subparts @xcite , _ i.e. _ , justifying the bound @xmath103 . actually , concerning
, it occurs that for all the structured sets listed above we have that either @xmath104 or both @xmath105 and @xmath106 have the same simplified closed - form upper bound , _
e.g. _ , they are both upper bounded by @xmath107 when @xmath99 .
thus , due to the observations made above , we will consider that @xmath108 can be bounded similarly to the actual gaussian mean width @xmath109 of the normalized set @xmath110 , _ i.e. _ , with the same simplified upper bound .
an example of this fact for the set of bounded @xmath33-sparse vectors is provided at the end of sec .
[ sec : proofs ] .
.useful properties of the gaussian mean width .
if not otherwise noted , all sets are subsets of @xmath14 .
@xmath111 : ( p[p : w - dialk ] ) is obtained by a simple use of the jensen and cauchy - schwartz inequalities , ( p[p : w - trans ] ) is a simple consequence of the triangular inequality and of @xmath112 .
[ cols= " < , < , < " , ] * paper organization : * the rest of the paper is structured as follows . in sec . [
sec : framework ] , we define the construction of our quantized sub - gaussian random mapping . additionally , this section characterizes the sub - gaussianity of its linear ingredient , _
i.e. _ , its random projection matrix , and its interplay with the `` anti - sparse '' nature of the mapped vectors . we also formalize and motivate the main objectives of the paper , _
e.g. _ , explaining the shape and the origins of the targeted quasi - isometric embedding with its two specific distortions .
[ sec : main - results ] provides the main results of this work , namely , _
( i ) _ the possibility to create with high probability a quasi - isometric sub - gaussian embedding from our quantized mapping ( prop .
[ prop : main - result ] ) , and _
( ii ) _ a study of this mapping s _ consistency width _ behavior ( prop .
[ prop : consistency - width ] ) .
[ sec : discussions ] discusses those two propositions , analyzing them in a few specific settings in comparison with related works in the fields of dimensionality reduction and 1-bit compressed sensing .
[ sec : nec - dither ] questions the necessity of dithering in the mapping @xmath113 and shows that , from an appropriate counterexample , our results do not hold in full generality without such a dither .
finally , sec .
[ sec : proofs ] and sec .
[ sec : proof - consistency - width ] contain the proofs of prop .
[ prop : main - result ] and prop .
[ prop : consistency - width ] , respectively , the auxiliary lemmas being demonstrated in appendix . [ [ conventions ] ] conventions : + + + + + + + + + + + + we find useful to summarize here our mathematical notations .
domain dimensions are denoted by capital roman letters , _
e.g. _ , @xmath114 vectors and matrices are associated to bold symbols , _
e.g. _ , @xmath22 or @xmath49 , while lowercase light letters are associated to scalar values .
the identity matrix in @xmath115 reads @xmath116 while @xmath117\in\{0,1\}$ ] is the indicator function of a set @xmath118 .
an `` event '' is a set whose definition depends on the realization of some random variables , _
e.g. _ , if @xmath119 is a random variable , the event @xmath120 has probability @xmath121 $ ] .
the @xmath122 component of a vector ( or of a vector function ) @xmath123 reads either @xmath124 or @xmath125 , and the vector @xmath126 may refer to the @xmath127 element of a set of vectors .
the set of indices in @xmath128 is @xmath129=\{1,\,\cdots , d\}$ ] .
the cardinality of a finite set @xmath130 reads @xmath131 . for any @xmath132 , the @xmath133-norm of @xmath123 is @xmath134 with @xmath135 .
the `` @xmath136-norm '' of a vector @xmath137 is @xmath138 , with @xmath139 the support of @xmath123 .
the @xmath140-sphere in @xmath141 is @xmath142 while the unit ball is denoted @xmath143 .
the diameter of a bounded set @xmath144 is written @xmath145 .
the set of @xmath33-sparse signals in @xmath14 is defined as @xmath146 while the set of @xmath33-sparse signals in an orthonormal basis ( onb ) @xmath147 , _ i.e. _ , with @xmath148 , reads @xmath149 .
the positive thresholding function is defined by @xmath150 for any @xmath151 . for @xmath152 , @xmath153 ( resp .
@xmath154 ) is the largest ( smallest ) integer smaller ( greater ) than @xmath155 .
a random matrix @xmath156 is a @xmath157 matrix with entries distributed as @xmath158 given the distribution parameters @xmath159 of @xmath160 ( _ e.g. _ , @xmath161 or @xmath162)$ ] ) .
a random vector in @xmath163 following @xmath164 is defined by @xmath165 . given two random variables @xmath166 and @xmath167 , the notation @xmath168 means that @xmath166 and @xmath167 have the same distribution .
since our developments do not focus on sharp bounds , we denote by @xmath169 or @xmath170 ( possibly large ) constants whose value can change between lines . in a few places , for simplicity , we write @xmath171 if there exists a constant @xmath57 such that @xmath172 , and correspondingly for @xmath173 .
moreover , @xmath174 means that @xmath171 and @xmath175 .
finally , for asymptotic relations , we use the common landau family of notations , _
i.e. _ , the symbols @xmath176 , @xmath177 and @xmath159 @xcite .
in this work , given a quantization resolution @xmath2 , we focus on the interaction between a random projection of @xmath14 into @xmath45 and the following uniform ( dithered ) quantizer , for some @xmath178 and @xmath179 , _
e.g. _ , for the quantizer mentioned in the introduction with @xmath180 and @xmath181 . ]
@xmath182 , applied componentwise on vectors in @xmath45 .
in other words , for some random matrix @xmath183 whose distribution is specified below , we study the properties of the mapping @xmath184 with @xmath185 where @xmath186)$ ] is a uniform _ dithering _ that stabilizes the action of @xmath54 @xcite .
we specialize the mapping on projection ( or sensing ) matrices @xmath24 with entries independently and identically drawn from a symmetric _ sub - gaussian _ distribution .
we recall that a random variable ( ) @xmath166 is sub - gaussian if its _ sub - gaussian norm _ ( or @xmath187-norm ) @xcite @xmath188 is finite . examples of sub - gaussian s are gaussian , bernoulli , uniform or bounded s , as @xmath189 sub - gaussian s are endowed with several interesting properties described , _
e.g. _ , in @xcite .
their tail is for instance bounded as the one of a gaussian , _
i.e. _ , there exists a @xmath57 such that for all @xmath190 and for a sub - gaussian @xmath166 , @xmath191 moreover , since @xmath192 , centering @xmath166 has no effect on its sub - gaussianity . by a slight abuse of notation ,
we denote collectively the distributions of _ symmetric _ sub - gaussian with zero expectation , unit variance and finite sub - gaussian norm @xmath193 by @xmath194 , with @xmath195 from .
this means that if @xmath196 , we do not fully specify the pdf of @xmath166 but we know that @xmath166 is centered , has unit variance and sub - gaussian norm @xmath193 . in this context , for a sub - gaussian random matrix @xmath197 , each row @xmath198 is also _ isotropic _ , _ i.e. _ , for all @xmath199 $ ] and all @xmath137 , @xmath200 however , conversely to the gaussian case where @xmath201 for @xmath202 and @xmath203 ( since @xmath204 ) , we do not necessarily have @xmath205 for @xmath206 and some absolute constant @xmath57 . as will be clear below
, we must anyway determine the deviations to this last equality .
interestingly , as noted in @xcite , any sub - gaussian random vector @xmath207 satisfies @xmath208 for some constant @xmath209 depending only the distribution of @xmath210 . while we have obviously @xmath211 if @xmath212 , it is possible to bound this constant in full generality .
indeed , up to a simple change of variable @xmath213 in the integral , is sustained by the berry - esseen central limit theorem ( as described in a simplified form in ( * ? ? ?
* theorem 4.2 ) ) .
this result shows basically that , for @xmath214 , the lhs of is bounded by @xmath215 for @xmath216 .
this means that @xmath217 for any @xmath210 .
notice , however , that this bound can be loose for many sub - gaussian distributions .
thanks to assumption , we can establish the behavior of the _ first absolute moment _
function @xmath218 since @xmath219 for any @xmath166 and using jensen s inequality , we indeed observe that @xmath220 @xmath221 for all @xmath137 . the last property , which is also considered in 1-bit cs with non - gaussian projections @xcite , is key for characterizing quantized embeddings from sub - gaussian projections .
having now fully described the elements composing our random quantized mapping @xmath222 , we formally address the objectives defined in the introduction by observing `` when '' , _
i.e. _ , under which conditions with respect to @xmath7 , there exist two small distortions @xmath223 such that the pseudo - distance @xmath224 is involved in the quasi - isometric relation @xmath225 for all pair of vectors taken in a general subset @xmath31 . in particular , we aim to control the distortions @xmath226 and @xmath227 with respect to @xmath7 , @xmath13 , the non - gaussian nature of @xmath24 ( _ i.e. _ , through @xmath193 and @xmath228 ) , the typical dimension of @xmath32 ( _ i.e. _ , its gaussian mean width ) and possible additional requirements on @xmath42 and @xmath43 .
let us justify and comment the specific form taken by .
first , @xmath229 is associated to a @xmath5-distance in the image of @xmath113 . as detailed in sec .
[ sec : proofs ] , this choice establishes an equivalence between the evaluation of @xmath229 and a specific counting procedure , _
i.e. _ , a count of the number of quantization _ thresholds _ separating each components of the randomly - projected vectors .
however , it is not clear if our developments can be extended to a @xmath6-based pseudo - distance , even if this holds , with additional distortion , in the case of gaussian random projections and for finite sets @xmath32 @xcite ( see sec . [
sec : discussions ] ) .
second , as explained in the introduction , a special case where both non - zero @xmath226 and @xmath227 appear specifies the constant @xmath230 in .
when @xmath231 , @xcite has proved a quantized version of the johnson lindenstrauss ( jl ) lemma showing that for a finite set @xmath232 of size @xmath233 , provided @xmath234 , one has @xmath235 for all pairs @xmath27 with a probability at least @xmath236 . as a direct impact of the loss of information induced by the quantization
, we also observe here that @xmath113 realizes a _ quasi - isometric _ mapping between @xmath237 and @xmath238 with @xmath239 and @xmath240 . finally , as will be clearly established in sec .
[ sec : quasi - isometric - embed ] , the anti - sparse nature of @xmath241 must be involved in the characterization of the right - hand side of in the case of a general sub - gaussian matrix @xmath24 .
indeed , let us consider a matrix with i.i.d .
bernoulli distributed random entries , _ i.e. _ , @xmath242 with @xmath243 for all @xmath244 and @xmath245 , the vectors @xmath246 and @xmath247 and assume @xmath248 , _
e.g. _ , with @xmath249 and @xmath250 .
then , taking @xmath251 , we clearly have @xmath252 and @xmath253 , so that @xmath254 and @xmath255
. consequently , if is expected to hold on any pair of vectors in @xmath32 , inserting @xmath42 and @xmath43 inside it gives @xmath256 . this limits our hope to have @xmath257 as small as we want by , _
e.g. _ , increasing @xmath7 .
in fact , between the two distortions , it is actually @xmath227 that should depend on the configuration of @xmath258 . as proved in app .
[ sec : absol - expect - dith ] , @xmath259).\ ] ] therefore , by definition of @xmath54 , from the independence of each component of @xmath222 and using the law of total expectation over @xmath260 and @xmath24 we have @xmath261 with @xmath210 and @xmath262)$ ] . from the assumption ( [ eq : one - moment - sg - antisparse ] ) and given @xmath263 , we then observe that @xmath264 for all vectors @xmath42 and @xmath43 such that @xmath241 belongs to the set @xmath265 this last set amounts to considering vectors that are not `` too sparse '' , _
i.e. _ , if @xmath266 then @xmath267 , which determines our notation @xmath268 as opposed to @xmath269
. however , the converse is not true and @xmath270 .
since belonging to @xmath268 prevents sparsity , we say that a vector @xmath271 is an _ anti - sparse _ vector of level @xmath272 .
actually states that , for vectors
@xmath273 , the expectation of @xmath274 is close to the one obtained with gaussian random projections , _
i.e. _ , close to the expectation @xmath275 associated to @xmath211 .
thus , if we expect to show that , for all vectors @xmath42 and @xmath43 in @xmath32 , @xmath274 concentrates around @xmath275 , we must take into account the anti - sparse nature of the difference @xmath241 , _
i.e. _ , we would need enforcing this vector to belong to @xmath268 for a sufficiently large @xmath276 . combining these three observations , and anticipating over the next section
, we can now refine the meaning of .
we are actually going to show that , if @xmath7 is bigger than some @xmath277 growing with the typical dimension of @xmath32 and decreasing with @xmath23 ( see sec . [
sec : main - results ] ) , then , with high probability , @xmath278 for all @xmath248 and @xmath279 . _ remark : _ as will be cleared later , our developments benefit of the tools and techniques developed in @xcite where it is shown that , for a 1-bit mapping @xmath280 such that @xmath281 with a random gaussian matrix @xmath282 , and for the normalized hamming distance @xmath283 $ ] , one has , provided @xmath284 and with probability exceeding @xmath236 , that for all @xmath285 , @xmath286 our extension to non - gaussian sensing matrices is also inspired by similar developments realized in @xcite for binary mappings and other generalized linear models .
in regards to the context explained in the previous section , our first main result can be stated as follows .
[ prop : main - result ] given @xmath2 , @xmath287 , @xmath288 , a bounded subset @xmath31 and a sub - gaussian distribution @xmath289 respecting for @xmath290 , there exist some values @xmath291 , only depending on @xmath193 , such that , if @xmath292 for a general set @xmath32 , or @xmath293 for structured sets @xmath32 ( see def .
[ def : structured - set ] for the definition of @xmath294 ) , such as the set of bounded @xmath33-sparse signals or the one of bounded rank-@xmath87 matrices , then , for @xmath295 , a dithering @xmath296)$ ] and the associated quantized mapping @xmath297 , we have with probability at least @xmath298 and for all pairs @xmath285 with @xmath279 , @xmath299 in the gaussian case , _
i.e. _ , for @xmath231 , the conditions remain the same and is simplified with @xmath300 , _
i.e. _ , there is no additional requirement on the anti - sparse nature of @xmath241 in since @xmath276 can be set to 1 and @xmath301 . in prop .
[ prop : main - result ] , as shown in sec .
[ sec : framework ] , the constant part @xmath302 of the multiplicative distortion appearing in both sides of is unavoidable in the case of non - gaussian projections ( with @xmath303 ) .
actually , we can show that this distortion can not decay faster than @xmath304 for non - gaussian ( but sub - gaussian ) random matrices when the level of anti - sparsity @xmath276 of @xmath305 increases . to see this ,
it is sufficient to study @xmath274 for an asymptotically large @xmath7 , _ i.e. _ , @xmath306 by the law of large numbers , and to observe how the relative error between @xmath306 and @xmath307 behaves when that level @xmath276 increases .
taking @xmath251 by simplicity , notice first that , from the observation made in , @xmath308 where @xmath309 and @xmath310 was introduced in .
let us then take @xmath42 and @xmath43 such that the vector @xmath311 is equal to 1 on its first @xmath276 components and zero elsewhere , _
i.e. _ , @xmath312 . in this case and
if @xmath24 is a random bernoulli matrix , @xmath313 is actually twice the _ mean absolute deviation _
( mad ) of a binomial distribution @xmath314 with @xmath276 degrees of freedom and success probability @xmath315 since @xmath316 and @xmath317 a bernoulli random variable such that @xmath318 , and @xmath319 .
however , from @xcite we can show that ( see app .
[ app : bin - bound ] for details ) @xmath320 for @xmath321 .
consequently , for our choice of @xmath322 such that @xmath323 , this shows that @xmath324 and proves that , even if we reached an asymptotic regime in @xmath7 , a multiplicative distortion between @xmath274 and @xmath325 would remain , and this one could decay faster than @xmath326 when @xmath276 increases .
it is therefore unclear if our decay in @xmath327 is optimal . to conclude this section ,
let us observe that prop .
[ prop : main - result ] improves a proof of existence of a quantized embedding given in ( * ? ? ?
* theorem 1.10 ) where it was showed that , provided @xmath328 , there exists an arrangement of @xmath7 affine hyperplanes in @xmath14 and a scaling factor @xmath329 such that @xmath330 where @xmath331 denotes the fraction of affine hyperplanes that separate the two vectors @xmath42 and @xmath43 . for reasons explained in sec .
[ sec : proofs ] , each element @xmath332 appearing in @xmath333 actually counts the number of parallel affine hyperplanes in @xmath14 normal to @xmath198 and far apart by @xmath41 , with a dithering that randomly displaces the origin .
therefore , prop .
[ prop : main - result ] basically constructs , in a random fashion , an arrangement of @xmath7 such parallel hyperplane bundle , _
i.e. _ , in @xmath7 different directions @xmath334\}$ ] . considering a gaussian matrix @xmath24 ( with @xmath300 ) , we have therefore proved that there with a minimal @xmath7 that grows like @xmath335 rather than @xmath336 when @xmath23 decays ( as expressed in ) .
this is even reduced to @xmath337 for pairs of vectors taken in a structured set . as a second important result
, we optimize the decay law ( as @xmath7 increases ) of the distance of any pair of vectors @xmath248 whose difference is `` not too sparse '' when those are mapped by @xmath222 on the same quantization point in @xmath338 , _
i.e. _ , when they are _
we refer to this distance as the _ consistency width _ of @xmath222 .
this width could be characterized from prop .
[ prop : main - result ] when @xmath339 , which provides @xmath340 ( or @xmath10 if @xmath32 is a structured set ) for large @xmath7 respecting ( resp . ) , @xmath41 fixed and @xmath341 small .
however , focusing on the conditions guaranteeing the consistency of @xmath42 and @xmath43 , and considering all quantities fixed but @xmath7 , our result below reaches the improved decay @xmath342 for a general set @xmath32 and @xmath343 for a structured one .
we prove the following proposition in sec .
[ sec : proof - consistency - width ] .
[ prop : consistency - width ] let us take a quantization resolution @xmath344 , an accuracy @xmath287 , a sub - gaussian distribution @xmath194 respecting for @xmath290 , @xmath345 such that @xmath346 and a bounded subset @xmath347 of @xmath14
. for a value @xmath57 depending only on @xmath193 , provided @xmath348 for a general set @xmath32 , or @xmath349 for a structured set @xmath32 , the map @xmath113 defined in with @xmath295 and @xmath350)$ ] is such that , with probability exceeding @xmath351 , @xmath352 for all @xmath353 with @xmath354 . in the gaussian case , _
i.e. _ , for @xmath231 , the conditions above remain the same with @xmath300 , _
i.e. _ , with no additional requirement on the anti - sparse nature of @xmath241 in . unfortunately , we were unable to produce a convincing counter example of a pair of vectors both with difference not in @xmath268 and failing to meet under the conditions of prop .
[ prop : consistency - width ] .
therefore , it is not clear if the condition @xmath273 is an artifact of the proof or if removing it could worsen then dependence in @xmath23 in .
before delving into the proofs of prop . [
prop : main - result ] and prop . [ prop : consistency - width ] ( see sec . [
sec : proofs ] and sec .
[ sec : proof - consistency - width ] , respectively ) , let us discuss their meaning and limitations , providing also some perspectives for future works . * on the impact of the diameter of structured sets : * for the structured sets considered in the introduction , it is known that if the linear embedding holds with high probability for all @xmath355 with some distortion @xmath17 , then , since is homogeneous , a simple rescaling argument proves that the same relation actually holds for all points in @xmath356 , or equivalently for all points in the cone @xmath84 if @xmath357 @xcite .
in particular , since such a linear embedding occurs with high probability for sub - gaussian random matrices provided @xmath358 @xcite , this requirement remains unchanged for reaching the embedding of vectors in @xmath84 .
obviously , in the case of a quantized embedding such as , the non - linear nature of @xmath54 prevents this rescaling argument from holding .
however , an interesting phenomenon occurs anyway in this case through the requirements and of prop .
[ prop : main - result ] and prop .
[ prop : consistency - width ] , respectively .
indeed , we see there that the diameter of the set @xmath32 has only a logarithmic impact on the minimal value of @xmath7 needed for these propositions to hold , since @xmath294 does not depend on the diameter of @xmath32 ( see def .
[ def : structured - set ] and the subsequent explanations ) .
this really slow increase approaches the scale - invariant requirement obtained by linear embedding of structured sets , and is anyway strikingly slower than the quadratic amplification of the minimal number of measurements provided by and in the case of a general set @xmath32 , as involved by ( p[p : w - hom ] ) when @xmath32 is expanded like @xmath359 for @xmath360 .
* mitigating the anti - sparsity requirement : * for both propositions , we can be concerned by the restriction that the vector difference must be `` not too sparse '' , _
i.e. _ , for @xmath353 there must be a sufficiently big @xmath276 , either for having @xmath361 and minimizing the distortion @xmath362 in , or for satisfying @xmath346 in prop .
[ prop : consistency - width ] .
however , in certain cases , it is possible to adapt the sensing matrix as to increase this @xmath276 . indeed , assuming without loss of generality that the vectors @xmath363 are expected to be `` too sparse '' only in @xmath364 when the sensing matrix is non - gaussian ( _ i.e. _ , @xmath303 ) , we can always `` rotate '' , @xmath365 is a rotation only if its determinant is @xmath366 . ]
@xmath32 with an onb @xmath367 of @xmath14 so that elements of @xmath368 with @xmath369 have a higher anti - sparse degree than those of @xmath370 , _
i.e. _ , @xmath371 possibly trying to maximize the left hand side in the selection of @xmath367 . therefore , while the requirements imposed on @xmath7 in prop .
[ prop : main - result ] and prop .
[ prop : consistency - width ] are unchanged between @xmath32 and @xmath84 in prop .
[ prop : main - result ] ( by the invariance ( p[p : w - invort ] ) of @xmath372 in table [ tab : gaussian - mean - width - prop ] ) and since @xmath373 for @xmath374 and @xmath375 , `` rotating '' @xmath32 with @xmath367 helps to lighten the condition imposed on @xmath258 .
moreover , this rotation is of course equivalent to directly build a sensing matrix @xmath376 to quasi - isometrically embed the set @xmath32 with the mapping @xmath377 . actually , in the case where @xmath378 as above , a good choice for @xmath367 is the dct basis , _
i.e. _ , using the incoherence of those two bases that prevents a sparse signal to be sparse in the frequency domain , also taking advantage of the fast fft - based matrix - vector multiplication offered by the dct . notice , however , that the procedure above can not work if @xmath32 is expected to generate differences of vectors that are sparse in different bases , _
e.g. _ , a union of incoherent bases such as @xmath379 and the dct basis .
in such a case , it could be hard to maximize the right - hand side of over @xmath367 .
interestingly , a similar procedure to the one described above has been developed recently in ( * ? ? ?
* theorem 2.3 ) in the context of fast circulant binary embeddings of finite sets of vectors .
the requirement on the anti - sparse nature of the mapped vectors is there mitigated by taking @xmath367 as the product of a hadamard transform with a diagonal matrix with random rademacher entries , which can provably reduce the _ coherence _
@xmath380 of too sparse @xmath123 with high probability .
* intrinsic `` anti - sparse '' distortion limit : * we can notice that for non - gaussian random measurements , the term @xmath381 in is actually lower bounded .
this is simply due to the relation @xmath382 , which implies @xmath383 whatever the properties of the vector @xmath384 .
consequently , @xmath385 which limits our hope to tighten the multiplicative error of quantized non - gaussian quasi - isometric embeddings , except if one considers asymptotic regimes where @xmath13 can be considered as being much larger than @xmath386 .
* distortion regimes : * as already noticed in @xcite , prop .
[ prop : main - result ] allows us to distinguish different regimes of the quasi - isometric embedding .
if @xmath387 , the quantization operator tends to the identity function and converges to a @xmath388 variant of the rip generalized to any sets @xmath32 and to sub - gaussian random matrices , as characterized in @xcite for general sets and in @xcite for sparse signal sets only . for @xmath389 the embedding becomes purely quasi - isometric and , keeping the context defined in prop .
[ prop : main - result ] , involves @xmath390 for some absolute constant @xmath57 .
however , in this case , the quantization becomes essentially binary .
in fact , it is exactly binary for random matrices whose entries are generated from a bounded symmetric sub - gaussian distribution , _
i.e. _ , from @xmath391 with @xmath392 for some @xmath393 . in this case , since @xmath32 is assumed bounded , for all @xmath394 , @xmath395 and the components of @xmath396 with @xmath397)$ ] can only take two values , _ e.g. _ , @xmath398 if @xmath399 . moreover , if @xmath400 is unbounded and @xmath399 , its sub - gaussian nature is so that the fraction of quantized measurements that do not belong to @xmath401 can be made arbitrarily close to 0 when @xmath41 increases . in conclusion , similarly to @xcite , we have basically defined a one - bit quantized embedding that preserves the norm of the projected vectors , as opposed to the mapping @xmath402 that loses this information @xcite .
notice there that the role of our dithering can be compared to the one of the threshold inserted in the sign quantization in @xcite .
conversely to that work , however , we do not provide any algorithm to reconstruct a signal from its quantized mapping by @xmath222 .
* towards an @xmath66 quasi - isometric embedding ?
* it is not clear if prop .
[ prop : main - result ] could be turned into a quasi - isometric embedding between @xmath52 and @xmath53 . as said earlier , for gaussian random matrices and for finite sets @xmath32
, an approximate quasi - isometric embedding can be found by integrating a non - linear distortion of the @xmath6-distance , _ i.e. _ , in for @xmath300 , @xmath403 is replaced by @xmath404 for some non - decreasing function @xmath405 .
interestingly , @xmath406 for @xmath407 and @xmath408 for @xmath409 , so that for small @xmath41 or large @xmath329 , @xmath410 . therefore , as soon as @xmath411 , we get approximately a @xmath66 quasi - isometric embedding .
knowing if this extends to any subset @xmath32 and to sub - gaussian random matrices is left for a future work . * reconstructing low - complexity vectors from quantized compressive observation ? * beyond the mere analysis of the quasi - isometric properties of our quantized mapping and closer to the context of quantized compressed sensing
, this paper does not say anything on the reconstruction algorithms that could be developed for recovering a signal @xmath42 from its observations @xmath412 . a few algorithms exist for realizing this operation , some when @xmath41 is small compared to the expected dynamic of @xmath413 @xcite , others in the 1-bit cs setting @xcite .
however , for the first category , their stability ( or convergence ) does not rely on a quasi - isometric embedding property but rather on the restricted isometry property @xcite or on variations involving other norms @xcite . in future research
, it will be appealing to find a proof of the instance optimality of those algorithms , _
e.g. _ , for the basis pursuit dequantizer ( bpdq ) , using the quasi - isometry property promoted by prop .
[ prop : main - result ] , even if recent interesting results show that an optimal `` non - rip '' proof can be developed for bpdq @xcite . *
extension to fast and universal quantized embeddings ?
* we conclude this section by mentioning that it would be useful to prove prop . [
prop : main - result ] for structured random matrices , _
e.g. _ , for random fourier or random hadamard ensembles @xcite , as recently obtained in @xcite for the binary embedding of finite sets .
this would lead to a fast computation of quantized mappings , with potential application in nearest - neighbor search for databases of high - dimensional signals .
an open question is also the possibility to extend this work to universally - quantized embeddings @xcite , _ i.e. _ , taking a periodic quantizer @xmath54 in .
this could potentially lead to quasi - isometric embeddings with ( exponentially ) decaying distortions on vectors sets with small gaussian width and using sub - gaussian random matrices .
considering the main results of this paper , namely prop . [
prop : main - result ] and prop .
[ prop : consistency - width ] , we could ask ourselves if a quantized mapping that would not include a dithering could also verify and under equivalent conditions on @xmath7 and on the anti - sparse nature of @xmath241 for any vectors @xmath414 in @xmath32 .
the answer is , however , negative in full generality , _ i.e. _ , it is possible to define a quantized and undithered map @xmath415 for some appropriate quantizer resolution @xmath41 and sub - gaussian random matrix @xmath24 that is _ incompatible _ with the definition of a quasi - isometric embedding with arbitrarily small additive distortion or with an arbitrarily small consistency width . to see this , let us set @xmath251 , @xmath416 ( applied componentwise defined in sec .
[ sec : framework ] .
we thus prefer to select @xmath54 as a rounding operation for the sake of clarity . ] ) , and take @xmath24 to be a bernoulli random matrix , _
i.e. _ , @xmath417
. given the value @xmath418 associated to the distribution of @xmath24 , we also set arbitrarily an integer @xmath276 such that @xmath419 .
in fact , we can compute that @xmath420 for a bernoulli , so that @xmath421 from the bound given in sec .
[ sec : framework ] .
therefore , @xmath422 certainly works .
we then define two @xmath276-sparse vectors @xmath423 with @xmath123 equal to 1 on it first @xmath276 components and 0 elsewhere , and @xmath424 for some fixed @xmath425 . clearly , when @xmath426 these two vectors belong to the structured set @xmath427 with @xmath428 .
moreover , from our definition of @xmath276 , the difference vector @xmath429 is adjustably `` anti - sparse '' since it lies in @xmath268 with @xmath430 .
interestingly , @xmath123 and @xmath431 are also consistent with respect to @xmath113 since @xmath432 .
this is due to the nature of quantization ( _ i.e. _ , a rounding to the closest integer ) and to the fact that both @xmath433 and @xmath434 .
let us now assume , as involved by prop .
[ prop : main - result ] , that for @xmath435 , it is possible to find @xmath7 arbitrarily large before @xmath436 so that , with high probability and for all @xmath248 with @xmath279 , @xmath437 with the constant @xmath57 defined in . however , by taking the consistent vectors @xmath438 and @xmath439 , this inequality leads by construction to @xmath440 in other words , since @xmath441 @xmath442 which is a clear contradiction .
we can similarly show that the same pair of consistent vectors @xmath438 and @xmath443 is incompatible with prop . [ prop : consistency - width ] as then the consistency width can not be arbitrarily small , even for asymptotically large @xmath7 . _ remark : _ interestingly , the counter - example above is easily hijacked to show that it is impossible for the un - dithered quantized mapping @xmath444 to respect the following property for an arbitrarily small @xmath17 and provided @xmath7 is large enough , @xmath445 where @xmath446 are some universal constants , @xmath447 is any positive function vanishing on equal inputs ( _ e.g. _ , a norm , a pseudo - norm or any metric ) and @xmath448 is any monotonically decreasing function with @xmath449 .
however , if @xmath54 is replaced by a sign operator as in @xcite , then the known binary @xmath23-stable embedding ( or b@xmath23se ) relates the _ angular _ distance between @xmath42 and @xmath43 to the hamming distance of their mappings , _
i.e. _ , two distances that are equal to zero in our counter - example above , which removes the contradiction . _
remark : _ the question whether dithering is necessary in the special case of a quantized mapping with a gaussian random matrix @xmath24 remains open .
the architecture of this proof is inspired by the one developed in @xcite for characterizing a 1-bit random mapping @xmath450 , @xmath451 . as will be clear below , some of the ingredients developed there had of course to be adapted to the specificities of @xmath113 and of our scalar quantization .
compared to @xcite we have also paid attention to optimize the dependency of @xmath7 to the desired level of distortions induced by @xmath113 in .
[ prop : main - result ] is proved as a special case of a more general proposition based on a `` softer '' variant of @xmath229 .
this new pseudo - distance is established as follows . defining the random mapping @xmath452 ,
with @xmath453 its @xmath122 component , we observe that for any @xmath454 , @xmath455,\ ] ] with the _ distinct sign event _ @xmath456 . in words
, for each @xmath199 $ ] , the sum over @xmath457 above simply counts the number of thresholds in @xmath458 separating @xmath459 and @xmath460 on the real line , since @xmath461 $ ] is equal to 1 for those and 0 for any other thresholds .
notice that the decomposition also justifies the observation made at the end of sec .
[ sec : consistency - width ] , namely the existence of uniform random tessellations of @xmath14 .
indeed , from the definition of @xmath113 , for each @xmath199 $ ] , @xmath462 $ ] also counts the number of parallel affine hyperplanes @xmath463 , all normal to @xmath198 and @xmath464 far apart , separating @xmath42 and @xmath465 .
in other words , @xmath14 is here tessellated with multiple so - called `` hyperplane wave partitions '' @xmath466\}$ ] @xcite with random orientations , periods and dithered origin . for @xmath467 .
_ on the top _ , @xmath468 and forbidden areas determined by @xmath469
are created when counting the number of thresholds @xmath470 separating @xmath471 and @xmath472 .
for instance , for an additional point @xmath473 as on the figure , @xmath474 but @xmath475 as @xmath476 lies in one forbidden area .
_ on the bottom _ figure , @xmath477 and threshold counting procedure operated by @xmath478 is relaxed .
now @xmath479 counts the number of limits ( in dashed ) of the green areas determined by @xmath469 , recording only one per thresholds @xmath470 , that separate @xmath471 and @xmath472 . here , for @xmath480 as on the figure , @xmath481 but @xmath482 .
] based on this observation , and as a generalization of an equivalent distance given in ( * ? ? ?
5 ) for binary mappings , we introduce for some @xmath483 the new pseudo - distance @xmath484,\ ] ] by defining the set @xmath485 the pseudo - distance @xmath486 is a non - increasing function of @xmath155 , with @xmath487 and @xmath488 the behavior of @xmath486 is best understood by introducing the one - dimensional distance @xmath489\ \in\ \delta { \mathbb}n,\quad { \rm for}\ a , b\in{\mathbb}r,\ ] ] so that @xmath490 fig .
[ fig : dt - behav ] explains how @xmath479 evolves for positive and negative @xmath155 , observing that , for each @xmath491 , @xmath492 determines forbidden or relaxed areas around the thresholds @xmath470 separating @xmath471 and @xmath472 and counted by @xmath479 . moreover , the next lemma , proved in app .
[ sec : proof - lemma - bounded - dbar ] , provides a first evaluation of the impact of the distance `` softening '' , by observing that , essentially , @xmath479 is not very far from both @xmath493 and @xmath494 for @xmath495 close to @xmath155 .
[ lem : bounded - dbar ] for any @xmath467 and @xmath496 , @xmath497 as announced above , we aim now at proving the next proposition whose special case @xmath498 leads to prop .
[ prop : main - result ] .
[ prop : main - relaxed - result ] given @xmath2 , @xmath287 , @xmath483 , @xmath499 , a bounded subset @xmath31 and a sub - gaussian distribution @xmath289 respecting for @xmath290 , there exist some values @xmath500 , only depending on @xmath193 , such that , if @xmath501 with @xmath502 the kolmogorov @xmath503-entropy of @xmath32 and the local set @xmath504 for @xmath505 , then for @xmath295 , a dithering @xmath296)$ ] , and the associated mapping @xmath113 defined in , we have with probability exceeding @xmath506 that for all pairs @xmath285 with @xmath279 , @xmath507 the proof sketch of prop . [ prop : main - relaxed - result ] is as follows : _ ( i ) _ given @xmath508 , we first show that the @xmath509 concentrates with high probability around @xmath325 up to a systematic bias @xmath510 due to the sub - gaussian nature of @xmath24 and controlled by the anti - sparse level of @xmath511 ; _ ( ii ) _ we take a finite covering of @xmath32 by a @xmath503-net @xmath512 ( for @xmath505 ) and we extend the concentration of @xmath509 to all vectors of @xmath513 by union bound ; _ ( iii ) _ we show that the softened pseudo - distance @xmath486 is sufficiently continuous in a neighborood of each pair of vectors in @xmath513 , which then allows us to extend to all pair of vectors in @xmath32 , as stated by prop .
[ prop : main - relaxed - result ] .
[ [ i - concentration - of - mathcaldtboldsymbolxboldsymboly ] ] _
( i ) concentration of @xmath509 : _
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + given a fixed pair @xmath514 , we show that @xmath509 concentrates around its mean by bounding its sub - gaussian norm as defined in . from , @xmath515 with the @xmath7 random variables @xmath516 for @xmath244 . however , the sum of @xmath517 independent sub - gaussian random variables @xmath518 is approximately invariant under rotation @xcite , which means that @xmath519 therefore , from , we find @xmath520 as shown in the following lemma ( proved in app .
[ sec : proof - lemma - subgaussian - dt ] by using lemma [ lem : bounded - dbar ] ) @xmath521 can be upper bounded ( and with it , the sub - gaussian norm of @xmath509 ) .
[ lem : subgaussian - dt ] let us take @xmath210 and @xmath262)$ ] .
for a fixed @xmath483 , the random variable @xmath522 is sub - gaussian with @xmath187-norm bounded by @xmath523 moreover , @xmath524 with @xmath525 if @xmath526 .
consequently , from and , @xmath527 is itself sub - gaussian with @xmath528 .
therefore , from the tail bound , there exists a @xmath57 such that for any @xmath81 @xmath529 { \leqslant}2 \exp\big(- c \epsilon^2 m\big).\ ] ] since @xmath530 and @xmath531 for all @xmath532 $ ] , provides @xmath533 for some constant @xmath534 , and @xmath535 { \leqslant}2 \exp\big(- c \epsilon^2 m\big).\ ] ] [ [ ii - extension - to - a - covering - of - mathcalk ] ] _ ( ii ) extension to a covering of @xmath32 : _ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + given a radius @xmath505 to be specified later , let @xmath513 an @xmath503-net of @xmath32 , _
i.e. _ , a finite vector set such that for any @xmath37 there exists a @xmath536 with @xmath537 . in particular ,
any vectors @xmath285 can then be written as @xmath538 for some @xmath539 and @xmath540 .
we also assume that the size of @xmath513 is minimal so that , by definition , @xmath541 , with @xmath542 the _ kolmogorov @xmath503-entropy _ of @xmath32 .
since there are no more than @xmath543 distinct pairs of vectors in @xmath513 , given @xmath483 , a standard union bound over shows that there exist some constant @xmath544 such that , if @xmath545 @xmath546\nonumber\\ \label{eq : concent - dt - cover - set } & \textstyle{\geqslant}\ 1\ , -\ , |{\mathcal}g_\eta|^2 \exp\big(- c\epsilon^2 m\big)\ { \geqslant}\ 1 - 2 \exp\big(- c''\epsilon^2 m\big).\end{aligned}\ ] ] [ [ iii - extension - to - mathbbrn - by - continuity - of - mathcaldt ] ] _ ( iii ) extension to @xmath14 by continuity of @xmath486 : _ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we can extend the event characterized in to all pairs of vectors in @xmath32 by analyzing the continuity property of @xmath486 in a limited neighborhood around the considered vectors .
we propose here to analyze this continuity with respect to @xmath6-perturbations of those vectors , as compared to @xmath5-perturbations in @xcite .
as will be clearer later , this allows us to reach a better control over @xmath7 with respect to @xmath23 .
[ lem : continuity - l2-dt ] let @xmath547 .
we assume that @xmath548 , @xmath549 for some @xmath550
. then for every @xmath483 and @xmath551 one has @xmath552 the proof is given in app .
[ sec : app - proof - cont - l2-perturb ] .
interestingly , the following proposition proved in app .
[ sec : proof - prop - small - resid - l2 ] shows that @xmath553 and @xmath554 can indeed be bounded uniformly for all @xmath555 .
[ lem : small - resid - l2 ] let @xmath556 be bounded , _
i.e. _ , @xmath557 and assume @xmath558 . then , for some @xmath57 , if @xmath559 for @xmath560 and with probability at least @xmath561 , we have for all @xmath562 @xmath563 _ i.e. _ , @xmath564 . for the sake of simplicity , we consider below the sub - gaussian parameter @xmath193 as fixed and integrate it in explicit or hidden constants , as in the notations `` @xmath565 '' or `` @xmath566 '' .
noting that @xmath567 and using a union bound over and , we get that if @xmath568 with probability higher than @xmath569 , for all @xmath539 and all @xmath570 , @xmath571 for some @xmath500 depending only on @xmath193 .
therefore , for any @xmath248 , using sequentially , , the upper bound given in lemma [ lem : continuity - l2-dt ] and provides @xmath572 however , given @xmath210 , using jensen s inequality , the reverse triangular inequality and , we find @xmath573 moreover , @xmath574 , so that , @xmath575 if @xmath279 , then induces @xmath576 and assuming @xmath577 , there exists a @xmath57 such that @xmath578 taking @xmath579 and @xmath580 , which gives @xmath581 and
@xmath582 , we find for another @xmath57 @xmath583 similarly , using , , the lower bound given in lemma [ lem : continuity - l2-dt ] and , we obtain @xmath584 finally , we have thus shown that there exist some @xmath291 such that for @xmath585 with probability at least @xmath586 the bound @xmath587 holds for all @xmath588 , which finishes the proof of prop .
[ prop : main - relaxed - result ] . as mentioned earlier , prop .
[ prop : main - result ] is thus obtained by simplifying the requirement appearing in prop . [
prop : main - relaxed - result ] .
first , for a general bounded set @xmath32 , since the sudakov inequality in ( p[p : w - suda ] ) provides @xmath589 , noticing that @xmath590 and that ( p[p : w - setinc ] ) and ( p[p : w - setdiff ] ) provide @xmath591 , we deduce that holds if @xmath592 as imposed in .
second , in the case of a quantized embedding of the structured sets defined in the introduction ( see def . [
def : structured - set ] ) , we can even reach a much weaker condition on @xmath7 . indeed , for such a set @xmath32 with @xmath593 , from and the definition of @xmath294
, we have for any @xmath594 @xmath595 so that , from , the right - hand side of can be bounded as @xmath596 this explains the simpler requirement needed for structured sets in prop .
[ prop : main - result ] .
_ example : _ let us conclude this section by deducing an upper bound on @xmath597 for the set @xmath598 ( with @xmath599 ) of bounded @xmath33-sparse vectors in an orthonormal basis @xmath600 of @xmath14 .
we first notice that since @xmath601 by invariance over the orthogonal group @xmath602 ( see ( p[p : w - invort ] ) in table [ tab : gaussian - mean - width - prop ] ) and from ( p[p : w - spars ] ) , @xmath603 moreover , the kolmogorov entropy is also invariant under @xmath602 , _
i.e. _ , @xmath604 and it is known that ( see , _
e.g. _ , @xcite ) @xmath605 by using stirling s bound .
this shows that @xmath606 with @xmath607 . additionally , since @xmath85 is invariant under dilation , @xmath608 and @xmath609 showing again , by matching with , that we have @xmath610 .
this confirms that @xmath105 has the same upper bound than @xmath106 .
therefore , for the structured set @xmath32 of bounded @xmath33-sparse vectors , ( and therefore ) is then satisfied if @xmath611
using the context defined in prop .
[ prop : consistency - width ] and for @xmath7 satisfying , we are going to show the contraposition of , _
i.e. _ , that with probability at least @xmath612 for some @xmath57 and for all @xmath285 with @xmath613 , having @xmath614 involves @xmath615 , or equivalently that @xmath616 from the definition of @xmath229 in .
the proof sketch is a follows .
first , for some @xmath505 , we create a finite @xmath503-covering of the set @xmath617 of vector pairs whose difference belongs to @xmath268 .
second , in order to show , we leverage the continuity of the pseudo - distance @xmath486 under @xmath6-perturbations ( lemma [ lem : continuity - l2-dt ] ) , as it happens that all points of @xmath618 are obtained by @xmath6-perturbations of the @xmath503-covering and that , moreover , those perturbations are stable under projections by @xmath24 ( lemma [ lem : small - resid - l2 ] ) . finally , we adjust @xmath503 and some additional parameters to show that , with high probability , the softened distance @xmath619 , for some @xmath155 depending on @xmath503 , is large enough over all pairs @xmath620 of the covering compatible with @xmath621 , hence inducing .
let us define the set @xmath622 .
we introduce a minimal @xmath503-net @xmath623 of @xmath618 with @xmath624 to be specified later , such that for all @xmath625 , there exists a @xmath626 with @xmath627 which also involves @xmath537 and @xmath628 .
the size of this minimal @xmath503-net is bounded as @xmath629 . indeed , by the semi - additivity of the kolmogorov entropy (
* theorem 2 ) , @xmath617 involves that @xmath630 for any @xmath631 .
since a @xmath632-net of @xmath633 can be obtained by the product @xmath634 , with @xmath635 and @xmath636 a @xmath637-net covering of @xmath32 , we obtain @xmath638 . as for the proof of prop .
[ prop : main - result ] in sec .
[ sec : proofs ] , by construction , all @xmath639 can also be written as @xmath640 with @xmath641 , @xmath642 .
notice that we have also @xmath643 , since @xmath644 and @xmath645 .
as stated by lemma [ lem : small - resid - l2 ] , the diameter of the _ local set _
@xmath646 is stable with respect to random projections . since @xmath567
, there exist indeed two values @xmath647 , only depending on the sub - gaussian norm @xmath193 , such that if @xmath648 and @xmath649 , we have with probability at least @xmath650 , @xmath651 therefore , @xmath652 and @xmath653 under the same conditions . moreover , if the previous event occurs , then , lemma [ lem : continuity - l2-dt ] for @xmath498 shows that for any @xmath551 , @xmath654 consequently , for reaching @xmath655 as expressed in , since @xmath656 involves @xmath657 , the proof can be deduced if we can guarantee that , for all @xmath658 with @xmath659 , the probability that @xmath660 tends ( exponentially ) to one with @xmath7 .
let us upper bound the corresponding probability of failure .
we can first observe the following result on a fixed pair of vectors .
this one is proved in app .
[ sec : proof - lemma - prob - bound - qdist - gtr - r ] .
[ lem : prob - bound - qdist - gtr - r ] let @xmath661 be in @xmath14 with @xmath662 for some @xmath345 and @xmath663 for @xmath664 . for @xmath40 , @xmath468 , @xmath665 $ ] , @xmath295 , @xmath666)$ ] and the pseudo - distance @xmath486 defined in , we have @xmath667 { \leqslant}\exp(-\frac{(mp - r)^2}{2mp}),\ ] ] with @xmath668 $ ] , @xmath669 and @xmath262)$ ] . moreover , if @xmath670 , @xmath671 from the discrete nature of @xmath486 , the previous lemma ( with @xmath155 set to @xmath672 ) shows that for a fixed pair of vectors @xmath673 holds with probability at least @xmath674 . moreover , if @xmath675 we have @xmath676 therefore , setting @xmath677 , gives @xmath678 { \geqslant}1 - \exp(-\frac{(mp - r)^2}{2mp } ) > 1 - 2\exp(-\frac{mp}{8}),\ ] ] if , from , @xmath679 thus , we have to adjust @xmath680 and @xmath503 in order to satisfy .
noting that @xmath681 if @xmath682 , _ i.e. _ , that we can set @xmath683 in lemma [ lem : prob - bound - qdist - gtr - r ] , this adjustment can be done from by imposing @xmath684 in @xmath685 a solution is to set , for some @xmath686 and @xmath93 to be specified later , @xmath687 and @xmath688 . then @xmath689 so that @xmath690 fixing @xmath691 and @xmath692 , a few estimations show finally that @xmath693 proving that for our choice of parameters , _
i.e. _ , for @xmath694 and @xmath695 , can be satisfied since @xmath684 . moreover , for this choice of parameters , provides @xmath696 we are now ready to complete the proof . using the previous developments , defining @xmath697 with @xmath698 fixed as above and @xmath699 as explained before , by a simple union bound there exist some constants @xmath500 such that if @xmath700 then the event @xmath701 holds with probability at least @xmath702 remembering that for having the diameter of @xmath646 must remain small under random projections by @xmath24 ( as stated in ) , so by imposing , we find again by union bound that for some other constants @xmath500 , if @xmath703 then , with probability at least @xmath704 , for all @xmath285 with @xmath273 and @xmath621 , combined with provides @xmath705 as requested at the beginning .
we conclude the proof by simplifying the general condition .
first , for a general bounded set @xmath32 , sudakov inequality ( p. [ p : w - suda ] ) and sec .
[ sec : proofs ] provide @xmath706 and @xmath707 , so that holds if @xmath708 for another constant @xmath21 .
second , if the set @xmath32 is structured , then , from and the same simplifications used for prop .
[ prop : main - relaxed - result ] to reach prop .
[ prop : main - result ] , the right - hand side of can be bounded by @xmath709 with @xmath710 , which explains the requirement .
we wish to gladly thank holger rauhut and sjoerd dirksen for interesting and enlightening discussion on quantized random projections during a short stay end of january 2015 in rwth aachen university , and jerry veeh ( auburn university , al , usa ) for interesting discussions on the error bounds of the stirling s approximation , as deduced in his lecture notes @xcite , and for having pointed out the work @xcite .
we also thank valerio cambareri ( uclouvain , belgium ) for interesting discussions on quantized embeddings and for his advices on the writing of this paper .
this short appendix proves the equality @xmath711).\ ] ] denoting @xmath712 , @xmath713 , @xmath714 and @xmath715 , since @xmath716 for any @xmath717 and @xmath718 , we can always write @xmath719 with @xmath720 . without loss of generality , we can assume that the @xmath166 is positive , _
i.e. _ , @xmath721 ( just flip the role of @xmath722 and @xmath723 if this is not the case ) . moreover , since @xmath724 , @xmath725 and @xmath726 therefore , @xmath727 if @xmath728 , then @xmath729 .
let us consider now the case @xmath730 .
if @xmath731 , then @xmath732 since @xmath733 , _ i.e. _ , @xmath734 since @xmath735 . consequently , provides @xmath736 . when @xmath737 , @xmath738 , _
i.e. _ , @xmath739 , and we get @xmath740 . in summary ,
@xmath741 in all cases , which proves the result .
we start by observing that @xmath742 - { { \mathbb}i}[{{\mathcal}f^s(a - k\delta , b - k\delta)}]\big|\\ & \textstyle { \leqslant}\sum_{k\in{\mathbb}z } { { \mathbb}i}[{{\mathcal}h^{t , s}(a - k\delta , b - k\delta)}]\end{aligned}\ ] ] with @xmath743 for @xmath744 , @xmath745 and @xmath746 , while for @xmath747 , @xmath748 .
moreover , a careful piecewise analysis made on the different sign combinations for @xmath495 and @xmath155 show that @xmath749\ } \cup \{|b| \in [ r_-,r_+]\}$ ] with @xmath750 and @xmath751 equals to @xmath752 if @xmath753 and 0 otherwise .
consequently , writing @xmath754 , @xmath755\ } \cup \{|b - k\delta|\in [ r_-,r_+]\}\big]\\ & { \leqslant}\ 2 \delta ( \tfrac{2r}{\delta } + 2 ) = 4 ( |t - s| + \delta ) .
\end{aligned}\ ] ] moreover , if @xmath756 , since then @xmath757 and @xmath758 , @xmath759 { \leqslant}2 \delta ( \tfrac{2|t|}{\delta } + 1 ) = 4|t| + 2\delta,\ ] ] and we find @xmath760
let us define @xmath761 with the two s @xmath762 and @xmath763 . from , @xmath764 .
moreover , from the approximate rotational invariance property , @xmath765 is sub - gaussian with @xmath766 , and using lemma [ lem : bounded - dbar ] and the bound @xmath767 , we find @xmath768 which demonstrates the sub - gaussianity of @xmath769 . for the expectation , writing @xmath770 and @xmath771 with @xmath772 and @xmath773 , by jensen s inequality and the law of total expectation , we find @xmath774 however , reusing some elements of the proof of lemma [ lem : bounded - dbar ] and considering @xmath775 fixed , @xmath776\\ & \textstyle{\leqslant}\delta\,\sum_{k\in{\mathbb}z } { \mathbb}e_\xi{{\mathbb}i}\big[\{|a'+\xi - k\delta|{\leqslant}|t|\}\big ] + \delta\,\sum_{k\in{\mathbb}z } { \mathbb}e_\xi{{\mathbb}i}\big[\{|b'+\xi - k\delta|{\leqslant}|t|\}\big ] .
\end{aligned}\ ] ] moreover , since @xmath262)$ ] , @xmath777&\textstyle=\sum_{k\in{\mathbb}z}\int_0^\delta{{\mathbb}i}\big[\{|a'+s - k\delta|{\leqslant}|t|\}\big]\,{\mathrm{d}}s\\ & = \textstyle\int_{{\mathbb}r } { { \mathbb}i}\big[\{|a'+s|{\leqslant}|t|\}\big ] { \mathrm{d}}s\ = \ 2|t| , \end{aligned}\ ] ] which provides also @xmath778 = 2|t|$ ] .
consequently , since these two quantities do not depend on @xmath775 , we find @xmath779 , @xmath780 , and @xmath781 .
we adapt the proof of lemma 5.5 in @xcite to both @xmath6-perturbations ( instead of @xmath5 ones ) of @xmath782 and @xmath783 , and to the context of uniform dithered quantization instead of 1-bit ( sign ) quantization . by assumption
, we have @xmath548 and @xmath549 .
therefore , the set @xmath784 : |({\boldsymbol}\phi { \boldsymbol}x')_i| { \leqslant}\eta \sqrt p , |({\boldsymbol}\phi { \boldsymbol}y')_i| { \leqslant}\eta \sqrt p\}\ ] ] is such that @xmath785 as @xmath786 . considering the definition of @xmath469 in , we have , for all @xmath787 and any @xmath717 , @xmath788 with @xmath789 . denoting @xmath790
, we find @xmath791\\ & { \leqslant}\textstyle \tfrac{\delta}{m } \sum_{i\in t } \sum_{k\in{\mathbb}z } { { \mathbb}i}[{\mathcal}f_i^{t}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta ) ] + \tfrac{\delta}{m } \sum_{i\in t^{{\tt c } } } \sum_{k\in{\mathbb}z } { { \mathbb}i}[{\mathcal}f_i^{t+\eta\sqrt p - a_i}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta)]\\ & { \leqslant}\textstyle \tfrac{\delta}{m } \sum_{i\in t } \sum_{k\in{\mathbb}z } { { \mathbb}i}[{\mathcal}f_i^{t}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta ) ] + \tfrac{\delta}{m } \sum_{i\in t^{{\tt c } } } \sum_{k\in{\mathbb}z } { { \mathbb}i}[{\mathcal}f_i^{t}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta)]\\ & \qquad\textstyle + \tfrac{1}{m } \sum_{i\in t^{{\tt c } } } \delta\sum_{k\in{\mathbb}z } \big|{{\mathbb}i}[{\mathcal}f_i^{t+\eta\sqrt p - a_i}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta ) ] - { { \mathbb}i}[{\mathcal}f_i^{t}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta)]\big|.\end{aligned}\ ] ] using to bound the last sum of the last expression and since , by definition of @xmath792 , @xmath793 for @xmath794 , we find @xmath795 however , @xmath796 and since @xmath797 for all @xmath152 , we find @xmath798 which provides the lower bound of . for the upper bound , @xmath799\\ & { \geqslant}\textstyle \tfrac{\delta}{m } \sum_{i\in t } \sum_{k\in{\mathbb}z } { { \mathbb}i}[{\mathcal}f_i^{t}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta ) ] + \tfrac{\delta}{m } \sum_{i\in t^{{\tt c } } } \sum_{k\in{\mathbb}z } { { \mathbb}i}[{\mathcal}f_i^{t-\eta\sqrt p+a_i}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta)]\\ & { \geqslant}\textstyle { { \mathcal}d}^{t}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y')\\ & \qquad\textstyle - \tfrac{1}{m } \sum_{i\in t^{{\tt c } } } \delta\sum_{k\in{\mathbb}z } \big|{{\mathbb}i}[{\mathcal}f_i^{t}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta ) ] - { { \mathbb}i}[{\mathcal}f_i^{t-\eta\sqrt p+a_i}({\boldsymbol}x_0 + { \boldsymbol}x',{\boldsymbol}y_0 + { \boldsymbol}y ' , k\delta)]\big|,\end{aligned}\ ] ] and , as above , the last sum can be upper - bounded by @xmath800 using .
we use here a similar proposition of mendelson with @xmath801 @xcite . ] _ et al . _ in @xcite for subsets of @xmath802 that we lift to subsets of @xmath803 thank to some tools developed in @xcite for other purposes .
we fix @xmath804 and form the set @xmath805 with @xmath806 . as @xmath807 , we know from ( * ? ? ?
* theorem 2.1 ) that for @xmath808 , @xmath809 and @xmath810 , @xmath811 { \leqslant}\exp(-c \tfrac{\epsilon^2 m}{\alpha^4}).\ ] ] however , for @xmath812 and @xmath813 , as observed similarly in @xcite , @xmath814 since , for all @xmath815 , @xmath816 , _
i.e. _ , @xmath817 .
therefore , fixing @xmath818 , if @xmath819 , with probability at least @xmath820 , we have , for all @xmath815 , @xmath821 where @xmath822 , @xmath823 is the last column of @xmath824 and using the fact that @xmath825 since @xmath826 .
therefore , replacing @xmath155 by its value , we find with the same probability , @xmath827 for all @xmath815 , _
i.e. _ , @xmath828 .
from the relation @xmath829 established in sec . [ sec : proofs ] between @xmath486 and @xmath830 defined in , and associated to the vectorial mapping @xmath831 whose components are independent , we reach the bound with the cdf of a binomial distribution : since @xmath832{\leqslant}&\textstyle\ { \mathbb}p\big[\,\big|\{j \in [ m]:\ , d^t({\phi^{^{_{\!\xi}}\!}}_i({\boldsymbol}u ) , { \phi^{^{_{\!\xi}}\!}}_i({\boldsymbol}v ) ) \neq 0\}\big| { \leqslant}r\big]\\ = & \textstyle\ \sum_{k=0}^{r } { m\choose k } p^k ( 1-p)^{m - k},\end{aligned}\ ] ] chernoff s inequality can upper bound this binomial cdf with @xmath833 { \leqslant}\exp(-\frac{(mp - r)^2}{2mp}).\ ] ] let us now lower bound @xmath834 . defining @xmath835 and @xmath836 , the action of dithering @xmath837)$ ] allows us to compute easily that , @xmath838 = { \mathbb}e \min\big(1 , \delta^{-1 } ( |{\boldsymbol}\varphi^{\top}{{\boldsymbol}w}|-2t)_+\big).\ ] ] in order to avoid any further singularity when @xmath839 , we can benefit from the fact that @xmath132 and work with this slightly looser bound : @xmath840 moreover , with @xmath841 , @xmath842 so that @xmath843 where @xmath844 and @xmath845 .
we can upper bound @xmath846 from our assumptions on the sub - gaussian vector @xmath210 : @xmath847 where the last inequalities rely on assumption ( setting @xmath848 ) and on the fact that @xmath312 .
moreover , for lower - bounding @xmath849 in , we observe that @xmath850 for @xmath851 .
therefore , defining @xmath852 and integrating by parts , we find @xmath853 where in the last inequality we used jensen s inequality and the convexity of @xmath854 .
it is easy to see that @xmath855 so that @xmath856 finally , @xmath857 the last expression providing if @xmath858 .
this small section establishes a lower bound on the approximation error of the mad @xmath859 of a binomial random variable @xmath860 by a fraction of its standard deviation @xmath861 when @xmath862 increases is well known ( see _ e.g. _ , @xcite ) .
specifically , we want to prove that @xmath863 for some absolute constant @xmath21 and all @xmath864 .
we start from the stirling s approximation of the factorial with an error bound due to r. w. gosper @xcite and redeveloped more clearly in @xcite ( see also @xcite for a similar bound ) : @xmath865 however , de moivre gave the following exact formula for @xmath866 @xcite , @xmath867 therefore , applying on this formula and using @xmath868 for @xmath869 , we find for @xmath864 @xmath870 or equivalently @xmath871 with @xmath321 , which provides the result .
m. a. davenport , j. n. laska , p. t. boufounos , and r. g. baraniuk . a simple proof that random matrices are democratic . technical report , rice university ece department technical report tree-0906 , houston , tx , november 2009 .
s. rane , p. t. boufounos , and a. vetro . quantized embeddings : an efficient and universal nearest neighbor method for cloud - based image retrieval . in _ proc .
spie applications of digital image processing xxxvi _ , san diego , ca , august 25 - 19 2013 . | under which conditions and with which distortions can we preserve the pairwise - distances of low - complexity vectors , _
e.g. _ , for _ structured sets _ such as the set of sparse vectors or the one of low - rank matrices , when these are mapped ( or embedded ) in a finite set of vectors ?
this work addresses this general question through the specific use of a quantized and dithered random linear mapping which combines , in the following order , a sub - gaussian random projection in @xmath0 of vectors in @xmath1 , a random translation , or _ dither _ , of the projected vectors and a uniform scalar quantizer of resolution @xmath2 applied componentwise .
thanks to this quantized mapping we are first able to show that , with high probability , an embedding of a bounded set @xmath3 in @xmath4 can be achieved when distances in the quantized and in the original domains are measured with the @xmath5- and @xmath6-norm , respectively , and provided the number of quantized observations @xmath7 is large before the square of the `` gaussian mean width '' of @xmath8 . in this case , we show that the embedding is actually _ quasi - isometric _ and only suffers of both multiplicative and additive distortions whose magnitudes decrease as @xmath9 for general sets , and as @xmath10 for structured set , when @xmath7 increases .
second , when one is only interested in characterizing the maximal distance separating two elements of @xmath8 mapped to the same quantized vector , _
i.e. _ , the `` consistency width '' of the mapping , we show that for a similar number of measurements and with high probability this width decays as @xmath11 for general sets and as @xmath12 for structured ones when @xmath7 increases . finally , as an important aspect of our work
, we also establish how the non - gaussianity of sub - gaussian random projections inserted in the quantized mapping ( _ e.g. _ , for bernoulli random matrices ) impacts the class of vectors that can be embedded or whose consistency width provably decays when @xmath7 increases . |
consider a homogeneous ideal @xmath0 $ ] , and let @xmath1 be homogenous polynomials in @xmath2 . since @xmath3 , we have @xmath4 . the @xmath5 of the variety of @xmath6 with respect to the variety of @xmath2 is defined as the quasiprojective variety @xmath7 .
we define the @xmath8 @xmath9 of an ideal @xmath2 to be the number of solutions in @xmath7 .
excess intersections are a well studied problem with applications in enumerative geometry , machine learning @xcite , and algebraic statistics @xcite . in addition
, there is a well developed theory of segre classes to study this problem that has been exploited in @xcite using computational algebraic geometry as well .
recent work by paolo aluffi has pushed this area even further in @xcite .
however , the motivation for this paper came at the 2012 institute for mathematics and its applications participating institution summer program for graduate students in algebraic geometry for applications by mike stillman .
we will focus on the numerical algebraic geometry perspective , where it is ideal to solve @xmath10 systems of equations , meaning the number of unknowns equals the number of equations .
so by understanding the zero - dimensional solutions of an excess intersection of an ideal , we can study the ideal itself .
our computations were performed with @xmath11 , @xmath12 , and @xmath13 .
we begin our study in the case that @xmath2 is an ideal generated by @xmath14 , and @xmath1 define a @xmath15 of equations of degree @xmath16 .
[ defbsys ] let @xmath2 be an ideal of @xmath17 $ ] generated by @xmath18 whose respective degrees are @xmath19 .
suppose @xmath16 is such that @xmath20 let @xmath21 denote a form of degree @xmath22 .
if the forms @xmath1 are given by @xmath23=\left[\begin{array}{lll } a_{11 } & \cdots & a_{1l}\\ a_{21 } & \cdots & a_{2l}\\ \vdots & & \vdots\\ a_{n1 } & \cdots & a_{nl } \end{array}\right]\left[\begin{array}{l } b_{1}\\ \vdots\\ b_{l } \end{array}\right],\ ] ]
then we say @xmath1 are a @xmath15 of degree @xmath16 .
the space of @xmath15 s with degree @xmath16 is parameterized by the coefficients of the homogeneous polynomials @xmath21 .
if @xmath2 is generated by @xmath24 , then we denote the excess number of a general @xmath15 with degree @xmath16 as @xmath25 . in the first section we will be interested in determining excess numbers of @xmath15 s where @xmath24 are monomials . at times
it will be more convenient to work with the equivalence number @xmath26 this definition is inspired by the notion of the _ equivalence of an ideal _ in @xcite [ chapter @xmath27 .
this number is the difference between the bezout bound and the excess number in the cases we consider .
the contributions of the paper include numerical algebraic geometry algorithms to compute excess numbers and a combinatorial proof of the theorem below .
this theorem can be proven easily using fulton - macpherson intersection theory , and in fact doing so generalizes the result to any ideal generated by a regular sequence .
but in the proof we present , we will see how @xmath25 and @xmath28 relate to the volume of a subdivided simplex .
the algorithms we present take advantage of the polyhedral structure in our problem to give bounds ( lower and upper - bound ) for an excess number .
[ thm : mainresult ] let @xmath2 be an ideal of @xmath17 $ ] generated by @xmath29 such that @xmath30 . if @xmath1 define a @xmath15 of degree @xmath16 , then @xmath31 where @xmath32 is the degree @xmath33 elementary symmetric function evaluated at @xmath34 and @xmath35 is the degree @xmath36 complete homogenous symmetric function evaluated at @xmath37 .
the paper is structured as follows .
we consider the case when @xmath2 is a monomial ideal , and show excess numbers equal mixed volumes of polytopes ( lemma @xmath38 ) . by further restricting to the case when the ideal @xmath2 defines a complete intersection that is also a linear space ( though not necessarily reduced ) , we do a mixed volume computation ( lemma @xmath39 ) to get an explicit formula for excess numbers . in the final section
, we present our algorithms that take advantage of the first sections results .
the author would like to thank alicia dickenstein for her many helpful comments to improve this paper as well as his advisor bernd sturmfels .
the key idea to theorem @xmath40 is to cast our excess intersection problem in the language of combinatorial geometry and prove lemma @xmath38 . since the following proofs will use newton polytopes , minkowski sums , and genericity , we set up additional notation here .
the @xmath41 of a form @xmath42 will be denoted as @xmath43 .
the standard @xmath44-simplex is the convex hull of the origin @xmath45 , and the standard basis of unit vectors @xmath46 in @xmath47 .
the binary operation , minkowski sum , will be denoted as `` @xmath48 '' .
now , we give examples of @xmath15 s .
[ trivalexample ] let @xmath49 be the trivial ideal of @xmath50 $ ] so that @xmath51 generates @xmath2 .
then , a @xmath15 of degree @xmath52 is given by @xmath53 homogenous polynomials @xmath54 @xmath55 here , @xmath56 is a polynomial of degree @xmath52 . for a general choice of @xmath57 ,
the excess number is @xmath52 .
when the coefficients of @xmath57 are specially chosen the excess number can decrease . in this case
, the excess number can only be less than @xmath58 if the discriminant of @xmath57 vanishes . whenever @xmath2 is a principal ideal , the excess numbers are easy to determine algebraically .
let @xmath59 be a principal ideal of the ring @xmath17 $ ] , and let @xmath60 .
then , a @xmath15 of degree @xmath16 is given by @xmath44 homogenous polynomials @xmath61 @xmath62 here , @xmath63 is a polynomial of degree @xmath64 . to determine the excess number @xmath65
, we saturate @xmath66 by @xmath2 .
doing so , we conclude that the excess intersection of @xmath66 is defined by @xmath67 and consists of finitely many points . by bezout s theorem , it follows that @xmath68 . when the @xmath69 are general polynomials of degree @xmath64 , the equality above holds .
[ montc ] let the ideal @xmath70 $ ] be generated by the forms @xmath71 then , a @xmath15 of degree @xmath72 is a system of @xmath73 quadrics which are linear combinations of @xmath74 .
a general @xmath15 in this case has four solutions not contained in @xmath75 .
so @xmath76 .
four is also the the mixed volume of the newton polytopes of @xmath77 . in lemma [ translateproblem ]
, we will see that this is not a coincidence .
[ tcfail ] let the ideal @xmath70 $ ] be generated by the forms @xmath78 then a @xmath15 of degree @xmath72 is a system of @xmath73 quadrics which are linear combinations of @xmath79 .
a general @xmath15 in this case is equal to the ideal @xmath2 and has no solutions outside of @xmath75 .
so @xmath80 .
in example [ montc ] the excess number was a mixed volume of the newton polytopes of @xmath81 , but in example [ tcfail ] , this was not the case .
now , we explain the differences between these two situations .
[ thm : translateproblem ] let @xmath0 $ ] be an ideal defined by the monomials @xmath24 .
if @xmath1 are a general @xmath15 of degree @xmath16 , then the excess number @xmath25 equals the mixed volume of the newton polytopes @xmath82 .
the set of @xmath15 s of degree @xmath16 is parameterized by the coefficients of @xmath21 in definition [ defbsys ] .
we denote the projective space associated to the coefficients of all of the polynomials @xmath21 by @xmath83 .
the dimension of this projective space is one less than the number of coefficients of all the @xmath21 .
we denote the projective space associated to the coefficients of the monomials of @xmath84 as @xmath85 .
so we have a natural map @xmath86 from @xmath83 to @xmath85 that maps the coefficients of @xmath21 to coefficients of @xmath87 .
important for our situation , is that when @xmath88 are monomials , the image of this map @xmath86 is onto .
let @xmath89 denote a zariski dense open subset of @xmath83 whose complement contains coefficients that give rise to @xmath15 s with excess numbers less than expected .
let @xmath90 be a zariski dense open set of coefficients of @xmath87 such that the mixed volume is less than expected .
if @xmath91 be in the inverse image of @xmath90 under @xmath86 , then the intersection of @xmath89 and @xmath91 is again a zariski dense open subset of @xmath83 . in particular , this means we can say a general @xmath15 s for which @xmath24 are monomials has @xmath25 equal to the mixed volume of @xmath82 .
in example [ montc ] , @xmath83 equals @xmath92 and @xmath85 equals @xmath93 . by a dimension count we see that @xmath86 is not onto and explains why the mixed volume can be different from the excess number . the key idea of this proof was the notion of a `` general @xmath15 ''
so that we could use bernstein s theorem to count the solutions we are interested in .
so to determine excess numbers of monomial ideals , we determine mixed volumes . in general , mixed volume computations are complicated , but in some cases there is hope for an explicit formula .
the case we consider is when @xmath94 is generated by powers of unknowns .
if @xmath1 define a @xmath15 of degree @xmath16 , then we determine the excess number @xmath65 by calculating the mixed volume of @xmath82 .
recall that the mixed volume @xcite [ chapter 8.5 ] can be calculated by determining the coefficient of @xmath95 in the polynomial defining the volume of the scaled minkowski sum @xmath96 if we let @xmath97 denote a simplex , then lemma @xmath98 shows that slicing @xmath97 by an appropriate hyperplane subdivides @xmath97 into two convex polytopes @xmath99 and @xmath100 .
the hyperplane can be chosen so that @xmath101 . since @xmath102 ,
we compute our desired mixed volume by determining the coefficients of @xmath103 in @xmath104 and @xmath105 ( lemma @xmath39 ) . to elucidate our ideas we consider the following example .
let @xmath106 be an ideal of the ring @xmath107 $ ] and let @xmath108 be a @xmath15 of degree @xmath109 .
then the newton polytope of @xmath110 will be the convex hull of two tetrahedra .
specifically , the newton polytope of @xmath110 is the convex hull of the following eight points in @xmath111 of which six are vertices of @xmath112 : @xmath113 to avoid confusion with points in @xmath114 , we describe points in @xmath111 as @xmath115 rather than @xmath116 .
we will also describe the newton polytope @xmath112 by its @xmath117 supporting hyperplanes rather than its vertices : @xmath118 the @xmath73 coordinate hyperplanes , @xmath118 the hyperplane defined by @xmath119 , and @xmath118 the hyperplane defined by @xmath120 .
+ the normal vectors of the @xmath117 hyperplanes supporting @xmath112 are the same for every @xmath121 .
indeed they are the standard unit vectors @xmath122 , the vector @xmath123 , and the vector @xmath124 . by standard polytope theory @xcite[proposition 7.12 ] , it follows that the scaled minkowski sum @xmath125 , has the same @xmath117 normal vectors as those of @xmath112 .
indeed , the supporting hyperplanes are @xmath118 the @xmath73 coordinate hyperplanes , @xmath118 the hyperplane defined by @xmath126 , and @xmath118 the hyperplane defined by @xmath127 .
+ now note that four of the five hyperplanes are defining facets of a simplex whose volume is @xmath128 .
the fifth hyperplane subdivides the simplex as seen in figure @xmath53 . by subtracting the volume of the white figure from the volume of the simplex ,
we attain the mixed volume by considering the coefficients of @xmath129 in the difference . in this example
, we would find the excess number satisfies @xmath130 .
we now precisely define the polytopes @xmath131 .
let @xmath132 and @xmath133 .
then @xmath97 is defined as the @xmath44-simplex whose @xmath134 vertices are the origin and @xmath135 . moreover
, the volume of @xmath97 equals @xmath136 which has a term @xmath137 when expanded out . if we define the hyperplane @xmath138 by @xmath139 , then @xmath138 slices @xmath97 into two convex polytopes @xmath99 containing the origin , and @xmath100 .
we will prove that @xmath100 is the scaled minkowski sum @xmath140 .
[ lem : simplexbounds ] let @xmath1 be a @xmath15 of degree @xmath16 .
if @xmath94 , then @xmath101 . we will show that the polytopes @xmath100 and @xmath140 have the same @xmath141 supporting hyperplanes so that they must be equal .
specifically , we show that the supporting hyperplanes are @xmath118 the @xmath44 coordinate hyperplanes , @xmath118 the hyperplane @xmath142 defined by @xmath143 , and @xmath118 the hyperplane @xmath138 defined by @xmath139 .
+ the first @xmath134 hyperplanes support the simplex @xmath97 .
the last hyperplane @xmath138 slices @xmath97 into two polytopes one containing the origin and a second which is by definition the polytope @xmath100 . now
because @xmath144 has @xmath141 supporting hyperplanes consisting of @xmath118 the @xmath44 coordinate hyperplanes , @xmath118 the hyperplane defined by @xmath145 , and @xmath118 the hyperplane defined by @xmath146 .
+ by standard polytope theory @xcite[proposition 7.12 ] , it follows @xmath140 has the desired supporting hyperplanes . with lemma @xmath98 , we are able to calculate the mixed volume by determining the coefficient of @xmath95 in an integral as seen in lemma @xmath39 . [
lem : finalintegral]with the previous notation , @xmath147 is a polynomial whose coefficient of @xmath95 equals by lemma @xmath98 , it follows that the volume of @xmath99 equals @xmath148d_{\triangle}\ ] ] with the bounds of each integral inside the brackets with respect to @xmath149 being @xmath150 and @xmath151 denotes the simplex in @xmath152-dimensional space with @xmath153 vertices of @xmath154 . by using the calculus fact @xmath155=\frac{1}{r!}\left(d - x_{k}-\cdots - x_{1}\right)^{r}$ ] and the binomial theorem
, we have @xmath156 with @xmath157 .
it is known how to integrate a linear form raised to some power over the simplex .
so to get the last equality , we use @xcite [ remark @xmath158 , that says @xmath159 . now , note that @xmath160 is congruent to @xmath161 modulo @xmath162 .
similarly , also note that @xmath163 is congruent to @xmath164 modulo @xmath162 .
so we have @xmath165 the last congruence is shown by an easy combinatorial argument .
we remark that the number @xmath65 depends only on the newton polytopes @xmath82 .
for example , consider the ideals @xmath166 , @xmath167 , @xmath168 in the ring @xmath169 $ ] .
all three of these ideals have the same excess numbers when every @xmath170 is greater than @xmath171 because the newton polytopes of the defining polynomials of a @xmath15 are the same for @xmath172 .
in particular , @xmath173 for @xmath172 .
but if we consider the ideal @xmath174 , we find the newton polytopes of a @xmath175 are different from those newton polytopes of a @xmath15 . in particular , one can compute the excess number @xmath176 to be @xmath177 .
we have given a combinatorial description of excess numbers of monomial ideals in the first part of the paper and used this idea to give an explicit formula in theorem @xmath40 .
in the last part of this paper , we give algorithms that use homotopy continuation , an idea from numerical algebraic geometry , to compute excess numbers of any ideal @xmath0 $ ] .
as mentioned in the introduction there are other ways to compute excess numbers with segre classes .
in addition , one can use off - the - shelf computer algebra software like ` macaulay2 ` to compute excess numbers by saturating the ideal of a @xmath15 by @xmath2 .
also , the examples we present here can also be worked out by hand using fulton - macpherson intersection theory .
our algorithms will construct two homotopies , called @xmath178 and @xmath179 , that take the isolated solutions of a @xmath180 @xmath181 as start points and tracks them to solutions of @xmath1 giving bounds on @xmath65 . in the first algorithm ,
the monomial ideal @xmath182 is constructed so that @xmath183 . by doing a numerical membership test @xcite [ chapter @xmath184
, we will determine @xmath65 and isolated solutions of @xmath7 explicitly .
in the second algorithm , the monomial ideal @xmath182 is constructed to give lower bounds of @xmath65 instead . but
by iterating the second algorithm , we have a probabilistic way to make this bound sharp and compute all isolated solutions of @xmath7 explicitly .
the @xmath178-homotopy gets its name because it produces an @xmath185er bound of @xmath65 prior to a membership test .
the @xmath179-homotopy gets its name because several @xmath186rations can produce sharp lower bounds of @xmath65 after a membership test .
let @xmath187 $ ] be forms such that @xmath188 with @xmath189 being a monomial multiplied by a scalar . to ease notation ,
let @xmath190 $ ] be a row vector whose entries sum to @xmath191 , @xmath192 be a row vector of @xmath193 different general forms , and @xmath194 be a row vector of a general form that is repeated @xmath193 times .
define the @xmath178-homotopy as @xmath195 @xmath196+\left(1-t\right)\left[\begin{array}{cccc } \overrightarrow{\beta_{1,1 } } & \overrightarrow{\beta_{1,2 } } & \cdots & \overrightarrow{\beta_{1,l}}\\ \vdots & & & \vdots\\ \overrightarrow{\beta_{n,1 } } & \overrightarrow{\beta_{n,2 } } & \cdots & \overrightarrow{\beta_{n , l } } \end{array}\right]\right)\left[\begin{array}{c } \overrightarrow{a_{1}}\mbox{}^{t}\\ \overrightarrow{a_{2}}\mbox{}^{t}\\ \vdots\\ \overrightarrow{a_{l}}\mbox{}^{t } \end{array}\right],\ ] ] with the degrees of the general forms of @xmath192 and @xmath194 chosen so that @xmath197 is a system of @xmath44 forms of degrees @xmath34 .
we denote the start points of @xmath178@xmath198 as @xmath199 and take them to be the isolated solutions of @xmath200@xmath201 . denote the end points of @xmath178@xmath198 as @xmath202 . with this definition , we have when @xmath203 that @xmath204 is a general @xmath180 @xmath181 of degrees @xmath16 . on the other hand , when @xmath205 we have @xmath204 is a @xmath15 of degree @xmath16 . by the fundamental theorem of parameter continuation of isolated roots @xcite [ theorem @xmath206 it follows that @xmath202 contains all isolated solutions of @xmath1 .
in particular , this proves theorem @xmath207 because @xmath208 .
now that we have the theorem , we present our algorithm .
+ + for this algorithm , we assume in step @xmath212 that the excess intersection of a monomial ideal can be determined .
we now give an example where @xmath2 defines the twisted cubic . [ exalg1 ] let the ideal @xmath70 $ ] be generated by the forms @xmath78 and suppose we want to calculate @xmath213 . to run the first algorithm , we input @xmath214 and @xmath215 . in step @xmath53 , we determine @xmath216 @xmath217+\left(1-t\right)\left[\begin{array}{cccccc } b_{11 } & b_{11 } & b_{12 } & b_{12 } & b_{13 } & b_{13}\\ b_{21 } & b_{21 } & b_{22 } & b_{22 } & b_{23 } & b_{23}\\ b_{31 } & b_{31 } & b_{32 } & b_{32 } & b_{33 } & b_{33 } \end{array}\right]\right)\left[\begin{array}{c } z^{2}\\ -yw\\ yz\\ -xw\\ y^{2}\\ -xz \end{array}\right].\ ] ] the forms @xmath218 and @xmath219 are general linear forms of @xmath220 $ ] .
once we have solved the system @xmath221 in step @xmath212 , we path track in step @xmath73 to calculate @xmath202 giving an upper bound @xmath222 of @xmath213 . in step @xmath171
, we use a numerical membership test @xcite to conclude @xmath223 . indeed , if @xmath224=\left[\begin{array}{cccccccc } 1/2 & 1 & 4/5 & 1/3 & 1/5 & 7/8 & 13 & 1/3\\ 3 & 7 & 9/7 & 1/8 & 4 & 1/6 & 5 & -1\\ -5 & 4 & 7/8 & 8/9 & 3 & 1/15 & 1/6 & -8\\ -1/4 & 2 & 1/3 & -1 & -1 & -2 & 7/9 & 1/4 \end{array}\right]^{t}\left[\begin{array}{c } x\\ y\\ z\\ w \end{array}\right],\ ] ] @xmath200then
we find the ten excess points are @xmath225 : let @xmath187 $ ] be forms that generate @xmath2 and @xmath228 be monomials that generate @xmath182 such that @xmath229 .
the @xmath179-homotopy is defined as @xmath230 @xmath231\left(t\left[\begin{array}{c } a_{1}\\ \vdots\\ a_{l } \end{array}\right]+\gamma\left(1-t\right)\left[\begin{array}{c } b_{1}\\ \vdots\\ b_{l } \end{array}\right]\right),\ ] ] with the degrees of the general forms @xmath21 equal to @xmath232 .
we denote the start points of @xmath179 as @xmath233 and take them to be the isolated solutions of @xmath200@xmath234 and denote the end points of @xmath235 as @xmath236 . with this definition , we have when @xmath237 equals @xmath53 that @xmath235 is a @xmath180 @xmath181 of degrees @xmath16 . on the other hand , when @xmath205 we have @xmath235 is a system of @xmath44 @xmath2-general forms of degrees @xmath34 .
while the @xmath179-homotopy is easy to set up , the fundamental theorem of parameter continuation of isolated roots @xcite [ theorem @xmath206 can not be applied .
so @xmath238 does not necessarily contain all isolated solutions of @xmath1 .
however , after doing a membership test , we can determine some points in @xmath238 are isolated solutions of @xmath1 .
so what we have is a lower bound on @xmath65 .
but , by iterating this homotopy , we can find more isolated solutions and give a better lower bound .
@xmath243\left(t\left[\begin{array}{c } z^{2}\\ yz\\ y^{2 } \end{array}\right]+\gamma\left(1-t\right)\left[\begin{array}{c } z^{2}-yw\\ yz - xw\\ y^{2}-xz \end{array}\right]\right)\ ] ] with @xmath21 the same as in example @xmath244 .
we find the @xmath245 isolated solutions of @xmath246 are @xmath247 : @xmath248 by taking @xmath239 to be different complex numbers and keeping @xmath219 fixed , with @xmath171 iterations , we were able to find that @xmath65 has a lower bound of @xmath249 .
by the previous subsection , we know that this lower bound is actually sharp
. we comment that with some choices of @xmath250 defining @xmath2 , it can happen that @xmath251 is greater than , equal to , or less than @xmath213 .
so one may be tracking too many paths , too few paths , or perhaps luckily the right number .
open questions remain about for which choice of monomials @xmath228 yield the best computational results .
in addition , how should we choose @xmath239 to find new solutions as we iterate ; and how can we verify that our lower bound has become sharp are also interesting questions .
these questions will remain for future work , and their answers may depend heavily on the context of the problem .
we remark that the @xmath179-homotopy need not have had the @xmath252 be monomials .
any choice of a form @xmath252 whose degree equals @xmath253 could have been used .
however , in this section , we have made the assumption that excess intersections of monomial ideals can be computed effectively , as we saw combinatorics can be used to understand excess numbers of monomial ideals . to conclude
, we have shown that determining excess numbers of monomial ideals can be reduced to computing a mixed volume in some cases .
with this idea , we are able to provide an explicit formula for excess numbers of ideals with general generators .
we presented two algorithms using numerical algebraic geometry to determine excess numbers of any ideal .
we also demonstrated that these algorithms have successfully lead to the calculation of excess numbers of an ideal defining the twisted cubic .
we believe that the the @xmath178-homotopy can compute excess numbers of many other ideals defined by sparse forms in many unknowns .
william fulton . , volume 2 of _ ergebnisse der mathematik und ihrer grenzgebiete .
3 . folge . a series of modern surveys in mathematics [ results in mathematics and related areas .
3rd series .
a series of modern surveys in mathematics]_. springer - verlag , berlin , second edition , 1998 . | we provide formulas and develop algorithms for computing the excess numbers of an ideal . the solution for monomial ideals
is given by the mixed volumes of polytopes .
these results enable us to design numerical algebraic geometry homotopies to compute excess numbers of any ideal . |
in @xcite ( see appendix for an authors version of this article ) , we proposed a maximum likelihood approach for blindly separating a linear - quadratic mixture defined by ( eq .
( 2 ) in @xcite ) : @xmath0 where @xmath1 and @xmath2 are two independent sources .
the log - likelihood for @xmath3 samples of the mixed signals @xmath4 and @xmath5 reads ( eq . ( 12 ) in @xcite ) : @xmath6+e_t[\log{f_{s_2}(s_2(t ) ) } ] -e_t[\log{|j(s_1(t),s_2(t))| } ] \label{finalcost_e.eq}\ ] ] where @xmath7 $ ] represents the time average operator on the @xmath3 samples , @xmath8 and @xmath9 are the probability density functions ( pdf ) of the sources @xmath1 and @xmath2 and @xmath10 is the jacobian of the mixture which reads ( eq . ( 4 ) in @xcite ) @xmath11 maximizing the log - likelihood requires that its gradient with respect to the parameter vector @xmath12 $ ] , _ i.e. _ @xmath13 , vanishes . defining the score functions of the two sources as ( eq .
( 13 ) in @xcite ) @xmath14 we can write ( eq . ( 14 ) in @xcite ) @xmath15-e_t[\psi_2(s_2)\frac{\partial s_2}{\partial { \bf w}}]-e_t[\frac{1}{j}\frac{\partial j}{\partial { \bf w } } ] \label{gradient_e.eq}\ ] ] rewriting ( [ mixture_model_e.eq ] ) in the vector form @xmath16 and considering @xmath17 as the independent variable and @xmath18 as the dependent variable , we can write , using implicit differentiation ( eq . (
15 ) in @xcite ) @xmath19 which yields ( eq . ( 16 ) in @xcite ) @xmath20 note that @xmath21 is the jacobian matrix of the mixing model . considering ( [ mixture_model_e.eq ] ) , we can write ( appendix in @xcite ) + @xmath22 and @xmath23 , which implies , from ( [ dsdw_e.eq ] ) @xmath24 and yields ( eq . ( 19 ) in @xcite ) @xmath25 $ } \nonumber \\ \frac{\partial s_2}{\partial { \bf w}}=\frac{1}{j}\mbox{\huge$[$}(l_2+q_2s_2)s_2 \;,\ ; ( 1-q_1s_2)s_1 \ ; , ( l_2+q_2s_2)s_1s_2 \;,\ ; ( 1-q_1s_2)s_1s_2 \mbox{\huge$]$ } \label{ds1s2dw_e.eq}\end{aligned}\ ] ] using ( [ ds1s2dw_e.eq ] ) , we obtain the first two terms of the gradient ( [ gradient_e.eq ] ) . to obtain the third term , we need to compute @xmath26 .
this partial derivative was computed inaccurately in @xcite so that equations ( 20 ) , and thus ( 17 ) , in @xcite are erroneous .
in @xcite we did not consider the implicit dependence of @xmath1 and @xmath2 on @xmath27 and computed the derivative of @xmath10 with respect to @xmath17 ignoring this dependence . considering @xmath28 ,
the correct equation for @xmath29 reads @xmath30 equation ( 20 ) in @xcite only corresponded to the first term on the right side of the above relation which reads , following ( [ jacobian_e.eq ] ) , as : @xmath31 $ } \label{djdw_scte_e.eq}\ ] ] we now compute the gradient ( [ djdw_e.eq ] ) entirely . considering ( [ jacobian_e.eq ] ) , we can write @xmath32 \label{djdws_e.eq}\ ] ] using ( [ djdw_e.eq ] ) , ( [ djdw_scte_e.eq ] ) , ( [ djdws_e.eq ] ) and ( [ dsdw_new_e.eq ] ) we finally obtain the following equation which must replace the equation ( 20 ) in @xcite @xmath33 \nonumber \\
\label{djdw_new_e.eq}\end{aligned}\ ] ] inserting ( [ ds1s2dw_e.eq ] ) and ( [ djdw_new_e.eq ] ) in ( [ gradient_e.eq ] ) , we obtain the following expression for the gradient which must replace equation ( 17 ) in @xcite @xmath34 $ } \nonumber \\ \label{gradient_new_e.eq}\end{aligned}\ ] ] 9 s. hosseini , y. deville , blind maximum likelihood separation of a linear - quadratic mixture , proceedings of the fifth international conference on independent component analysis and blind signal separation ( ica 2004 ) , pp .
694 - 701 , issn 0302 - 9743 , isbn 3 - 540 - 23056 - 4 , springer - verlag , vol .
lncs 3195 , granada , spain , sept .
22 - 24 , 2004 .
it is well known that the independence hypothesis is not sufficient for separating general nonlinear mixtures because of the very large indeterminacies which make the nonlinear bss problem ill - posed .
a natural idea for reducing the indeterminacies is to constrain the structure of mixing and separating models to belong to a certain set of transformations .
this supplementary constraint can be viewed as a regularization of the initially ill - posed problem . in this paper
, we study a linear - quadratic mixture model which may be considered as the simplest ( nonlinear ) version of a general polynomial model . our main aim is to develop an approach which can be easily extended to higher - order polynomial models .
hence , we propose a recurrent separating structure whose realization does not require the knowledge of the explicit form of the inverse of the mixing model .
we develop a rigorous method to identify the parameters of the separating structure in a maximum likelihood framework .
the algorithm is developed so that the inverse of the mixing structure is not required to be known .
thus , it can be extended to more general polynomial mixtures .
suppose @xmath35 and @xmath36 are two independent random signals . given the following nonlinear instantaneous mixture model @xmath37 we would like to estimate @xmath35 and @xmath36 up to a permutation and a scaling factor ( and possibly an additive constant )
. for simplicity , let s denote @xmath38 and @xmath39 .
@xmath1 and @xmath2 will be referred to as the _ sources _ in the following .
( [ firstmodel.eq ] ) can be rewritten as @xmath40 in which @xmath41 and @xmath42 represent the linear contributions of the sources in the mixture , and @xmath43 and @xmath44 represent the quadratic contributions .
the negative signs are chosen for simplifying the notations of the separating structure .
solving the model ( [ mixture_model.eq ] ) for @xmath1 and @xmath2 leads to the following two pairs of solutions , which may be considered as two direct separating structures : @xmath45 where @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 and @xmath52 .
it can be easily verified that @xmath53 , where @xmath10 is the jacobian of the mixing model ( [ mixture_model.eq ] ) and reads @xmath54 according to the variation domain of the two sources , three different cases may be considered : \1 ) @xmath55 for all the values of @xmath1 and @xmath2 . in this case
( [ inverse.eq ] ) becomes : @xmath56 @xmath57 thus , the first direct separating structure in ( [ inverse.eq ] ) leads to the actual sources and the second direct separating structure leads to another solution , equivalent to the first one up to a permutation , a scaling factor , and an additive constant .
\2 ) @xmath58 for all the values of @xmath1 and @xmath2 . in this case
, the first structure leads to the permuting solution , defined by ( [ pair2.eq ] ) , and the second structure to the actual sources @xmath59 .
an example is shown in fig .
[ case2.fig ] for the numerical values @xmath60 , @xmath61 , @xmath62 , @xmath63 and @xmath64 $ ] .
\3 ) @xmath58 for some values of the sources and @xmath55 for the other values . in this case
, each structure leads to the non - permuted sources ( [ pair1.eq ] ) for some values of the observations and to the permuted sources ( [ pair2.eq ] ) for the other values .
an example is shown in fig .
[ case3.fig ] ( with the same coefficients as in the second case , but for ) .
the permutation effect is clearly visible in the figure .
one may also remark that the straight line @xmath65 in the source plane is mapped to a conic section in the observation plane ( shown by asterisks ) .
thus , it is clear that the direct structures may be used for separating the sources if the jacobian of the mixing model is always negative or always positive , _
i.e. _ for all the source values .
otherwise , although the sources are separated _ sample by sample _
, each retrieved signal contains samples of the two sources .
this problem arises because the mixing model ( [ mixture_model.eq ] ) is not bijective .
this theoretically insoluble problem should not discourage us .
in fact , our final objective is to extend the idea developed in the current study to more general polynomial models which will be used to approximate the nonlinear mixtures encountered in the real world .
if these real - world nonlinear models are bijective , we can logically suppose that the coefficients of their polynomial approximations take values which make them bijective on the variation domains of the sources .
thus , in the following , we suppose that the sources and the mixture coefficients have numerical values ensuring that the jacobian @xmath10 of the mixing model has a constant sign .
the natural idea to separate the sources is to form a direct separating structure using any of the equations in ( [ inverse.eq ] ) , and to identify the parameters @xmath66 , @xmath67 , @xmath68 and @xmath69 by optimizing an independence measuring criterion .
although this approach may be used for our special mixing model ( [ mixture_model.eq ] ) , as soon as a more complicated polynomial model is considered , the solutions @xmath70 can no longer be determined so that the generalization of the method to arbitrary polynomial models seems impossible . to avoid this limitation , we propose a recurrent structure shown in fig .
[ model.fig ] .
note that , for @xmath71 , this structure is reduced to the basic hrault - jutten network .
it may be checked easily that , for fixed observations defined by ( [ mixture_model.eq ] ) , @xmath72 and @xmath73 corresponds to a steady state for the structure in figure [ model.fig ] .
the use of this recurrent structure is more promising because it can be easily generalized to arbitrary polynomial models .
however , the main problem with this structure is its stability .
in fact , even if the mixing model coefficients are exactly known , the computation of the structure outputs requires the realization of the following recurrent iterative model @xmath74 where a loop on @xmath75 is performed for each couple of observations @xmath76 until convergence is achieved .
it can be shown that this model is locally stable at the separating point @xmath77 , if and only if the absolute values of the two eigenvalues of the jacobian matrix of ( [ recurrent.eq ] ) are smaller than one . in the following ,
we suppose that this condition is satisfied .
let @xmath78 be the joint pdf of the sources , and assume that the mixing model is bijective so that the jacobian of the mixing model has a constant sign on the variation domain of the sources .
the joint pdf of the observations can be written as @xmath79 taking the logarithm of ( [ pdf1.eq ] ) , and considering the independence of the sources , we can write : @xmath80 given n samples of the mixtures @xmath81 and @xmath82 , we want to find the maximum likelihood estimator for the mixture parameters @xmath83 $ ] .
this estimator is obtained by maximizing the joint pdf of all the observations ( supposing that the parameters in @xmath17 are constant ) , which is equal to @xmath84 if @xmath85 and @xmath86 are two i.i.d .
sequences , @xmath87 and @xmath88 are also i.i.d . so that @xmath89 and @xmath90 .
the cost function to be maximized can be defined as @xmath91 , which will be denoted using the temporal averaging operator @xmath7 $ ] as @xmath92\ ] ] using ( [ logf1.eq ] ) : @xmath6+e_t[\log{f_{s_2}(s_2(t ) ) } ] -e_t[\log{|j(s_1(t),s_2(t))| } ] \label{finalcost.eq}\ ] ] maximizing this cost function requires that its gradient with respect to the parameter vector @xmath17 , _ i.e. _ @xmath13 , vanishes . defining the score functions of the two sources as @xmath14 and considering that @xmath93 , we can write @xmath15-e_t[\psi_2(s_2)\frac{\partial s_2}{\partial { \bf w}}]-e_t[\frac{1}{j}\frac{\partial j}{\partial { \bf w } } ] \label{gradient.eq}\ ] ] rewriting ( [ mixture_model.eq ] ) in the vector form @xmath16 and considering @xmath17 as the independent variable and @xmath18 as the dependent variable ,
we can write , using implicit differentiation @xmath19 which yields @xmath94 note that @xmath21 is the jacobian matrix of the mixing model . using ( [ gradient.eq ] ) and ( [ dsdw.eq ] ) , the gradient of the cost function @xmath95 with respect to the parameter vector @xmath17 is equal to ( see the appendix for the computation details ) @xmath96 $ } \label{dldw.eq}\end{aligned}\ ] ] in practice , the actual sources and their density functions are unknown and will be replaced by the reconstructed sources , _
i.e. _ by the outputs of the separating structure of fig [ model.fig ] , @xmath97 , in an iterative algorithm
. the score functions of the reconstructed sources can be estimated by any of the existing parametric or non - parametric methods . in our work
, we used a kernel estimator based on third - order cardinal splines . using ( [ dldw.eq ] ) ,
the cost function ( [ finalcost.eq ] ) can be maximized by a gradient ascent algorithm which updates the parameters by the rule @xmath98 .
the learning rate parameter @xmath99 must be chosen carefully to avoid the divergence of the algorithm .
note that the algorithm does not require the knowledge of the explicit inverse of the mixing model ( direct separating structures ( [ inverse.eq ] ) ) .
hence , it can be easily extended to more general polynomial mixing models . considering ( [ mixture_model.eq ] ) , we can write + @xmath22 and @xmath23 , which implies , from ( [ dsdw.eq ] ) @xmath100 which yields @xmath25 $ } \nonumber \\ \frac{\partial s_2}{\partial { \bf w}}=\frac{1}{j}\mbox{\huge$[$}(l_2+q_2s_2)s_2 \;,\ ; ( 1-q_1s_2)s_1 \ ; , ( l_2+q_2s_2)s_1s_2 \;,\ ; ( 1-q_1s_2)s_1s_2 \mbox{\huge$]$ } \label{ds1s2dw.eq}\end{aligned}\ ] ] considering ( [ jacobian.eq ] ) @xmath101 $ } \label{djdw.eq}\end{aligned}\ ] ] ( [ dldw.eq ] ) follows directly from ( [ gradient.eq ] ) , ( [ ds1s2dw.eq ] ) and ( [ djdw.eq ] ) . | an error occurred in the computation of a gradient in @xcite . the equations ( 20 ) in appendix and ( 17 ) in the text were not correct .
the current paper presents the correct version of these equations . |
the excitation of nucleon resonances in electromagnetic interactions has long been known as an important source of information on the baryon structure and long- and short - range interaction in the domain of quark confinement .
constituent quark models have been developed that relate electromagnetic resonance transition form factors to fundamental quantities , such as the quark confining potential .
while the picture of the @xmath3 and @xmath4 built from 3 quarks is recognized as a basic starting point in the description of these states , the inter - quark interaction does not exclude the possibility of additional degrees of freedom , namely , the presence of @xmath5 components in the @xmath3 and @xmath4 .
there is also a possibility of alternative structures , such as hybrid @xmath6 states and resonances dynamically generated in the meson - baryon interaction .
the @xmath7 dependence of the @xmath0 transition amplitudes is highly sensitive to different descriptions of the nucleon resonances .
rich information on these amplitudes , both transverse and longitudinal , has become available in a wide @xmath7 range for the n(1440)p@xmath8 , n(1520)d@xmath9 , n(1535)s@xmath8 resonances due to the recent high precision jlab - clas measurements of the @xmath10 reactions @xcite .
more results are expected in @xmath11 and @xmath12 electroproduction .
while comparing the predictions obtained in different approaches with the @xmath0 amplitudes extracted from experimental data , we found that there are two points which introduce confusion in theoretical predictions and in most cases do not allow to compare properly the results of different approaches with each other and with experimental data : \(i ) it is known that experimental results on the @xmath0 helicity amplitudes @xmath13 , extracted from the contribution of the diagram of fig . 1 to @xmath14 , contain the sign of the @xmath15 vertex .
however , by the definition accepted many years ago , this fact is not reflected explicitly in the amplitudes extracted from experiment . in many cases
this is not taken into account in theoretical calculations and causes difficulties in dealing with the common sign of the predicted amplitudes .
\(ii ) the definition of the amplitudes @xmath13 through the hadron electromagnetic current which is commonly used in theoretical approaches contains points which can introduce mistakes in the relative sign of longitudinal amplitude @xmath2 relative to the transverse ones @xmath16 .
this is the second source of difficulties in dealing with theoretical predictions .
our goal in this paper is to present different definitions of the @xmath17 helicity amplitudes : through the @xmath18 multipole amplitudes , through the hadron electromagnetic current , in terms of the @xmath17 form factors , and in quark model , which are consistent with each other . to exclude possible sources of mistakes , we give explicitly the definitions of all quantities which enter the formulas .
we pay special attention to the way , how the sign of the @xmath15 vertex should be taken into account .
it is known that this sign comes from the relative contributions of the diagrams which present the resonance ( fig
. 1 ) and born terms ( fig .
2 ) contributions to @xmath14
. we will give formulas which relate the @xmath17 helicity amplitudes extracted from experiment to those which are calculated through the hadron electromagnetic current and are multiplied by the sign of the ratio of the coupling constants in the vertices @xmath19 and @xmath15 . .
, width=445 ] in sec .
ii , we present the definitions of the @xmath17 helicity amplitudes through the @xmath18 multipole amplitudes , which are used in the extraction of @xmath13 from the @xmath18 data .
these definitions are well known and coincide , for example , with the definitions presented in refs .
@xcite . in sec .
iii , we present the definition of the @xmath17 helicity amplitudes through the hadron electromagnetic current . to distinguish between these amplitudes and those extracted from experiment , we denote the amplitudes defined through the hadron electromagnetic current by @xmath20 : @xmath21 here @xmath22 is the sign which reflects the presence of the @xmath23 vertex in fig
. 1 . in sec .
iv , using the definitions of sec .
iii , we present the expressions for the amplitudes @xmath20 in terms of the @xmath24 form factors . in sec .
v , the relation between @xmath22 and the sign of the ratio of the @xmath25 , @xmath23 coupling constants is found from the covariant calculations of the figs .
1,2 contributions to @xmath14 . in sec .
vi , we present formulas for the calculation of the amplitudes @xmath20 in nonrelativistic quark model . to avoid possible sources of mistakes , we give explicitly the definitions of all quantities which enter these formulas and present in explicit form definitions of wave functions for the @xmath3 , @xmath26 , and @xmath27
further , in sec .
vi , we present formulas for the amplitudes @xmath28 for the transitions @xmath29 in nonrelativistic quark model . in sec .
vii , we present information on the signs of the @xmath15 coupling constants available in quark model , and demonstrate the importance of using these signs for the correct presentation of quark model predictions .
in other approaches , such as dynamically generated resonances , these signs should be found within these approaches . only in this case , we can make proper comparison of the predictions obtained in different approaches with each other and with the results extracted from experiment .
the @xmath24 helicity amplitudes , extracted from experimental data on the reaction @xmath30 , are usually presented through the @xmath14 multipole amplitudes . in the case of the @xmath17 amplitudes ,
the corresponding formulas are following . for @xmath31 multipole
amplitudes : @xmath32 , \\ & & a_{3/2}=\frac{\left[l(l+2)\right]^{1/2}}{2 } ( { \cal e}_{l+}-{\cal m}_{l+}),\\ & & s_{1/2}=-\frac{1}{\sqrt{2}}(l+1){\cal s}_{l+}.\end{aligned}\ ] ] for @xmath33 multipole amplitudes : @xmath34,\\ & & a_{3/2}=-\frac{\left[l(l+2)\right]^{1/2}}{2 } ( { \cal e}_{(l+1)-}+{\cal m}_{(l+1)-}),\\ & & s_{1/2}=-\frac{1}{\sqrt{2}}(l+1 ) { \cal s}_{(l+1)-}.\end{aligned}\ ] ] here @xmath35 are the amplitudes which are related to the resonance contributions to the multipole amplitudes at the resonance positions via : @xmath36^{1/2},\\ & & c_{1/2}=-\sqrt{\frac{1}{3}},~c_{3/2}=\sqrt{\frac{2}{3 } } ~for~\gamma^ * p\rightarrow \pi^0 p,\\ & & c_{1/2}=-\sqrt{\frac{2}{3}},~c_{3/2}=-\sqrt{\frac{1}{3 } } ~for~\gamma^ * p\rightarrow \pi^+ n.\end{aligned}\ ] ] in eq .
( 9 ) , @xmath37 , @xmath38 , @xmath39 and @xmath40 are the total width , mass , spin and isospin of the resonance , @xmath41 is its branching ratio to the @xmath42 channel , @xmath43 is the nucleon mass , @xmath44 and @xmath45 are momenta of the real photon and pion at the resonance position in the c.m.s . with this definition ,
the @xmath46 width is @xmath47 below we give the relations between multipole amplitudes @xmath48 , @xmath49 , @xmath50 and the @xmath51 differential cross section .
it is convenient to present these relations in terms of the intermediate amplitudes @xmath52 : @xmath53p'_{l-1}(x)\},\\ f_2=&&\sum[(l+1)m_{l+}+lm_{l- } ] p'_l(x ) , \\
f_3=&&\sum[(e_{l+}-m_{l+})p''_{l+1}(x)\nonumber\\ & & + ( e_{l-}+m_{l-})p''_{l-1}(x)],\\ f_4=&&\sum(m_{l+}-e_{l+}-m_{l-}-e_{l-})p''_l(x ) , \\ f_5=&&\sum[(l+1)s_{l+}p'_{l+1}(x)-
ls_{l-}p'_{l-1}(x)],\\ f_6=&&\sum[ls_{l-}-(l+1)s_{l+ } ] p'_l(x).\end{aligned}\ ] ] the amplitudes @xmath52 are related to the helicity amplitudes and the cross section of the @xmath51 reaction in the following way :
@xmath54 @xmath55 @xmath56.\end{aligned}\ ] ] in eqs.(13 - 29 ) , @xmath57 and @xmath58 are the momenta of the virtual photon and pion in the c.m.s . of the reaction @xmath51 , @xmath59 and @xmath60
are the polar and azimuthal angles of the pion , @xmath61 is the polarization factor of the virtual photon , @xmath62 .
the definition of the @xmath63 helicity amplitudes through the hadron electromagnetic current , which is commonly used for the calculation of these amplitudes in theoretical approaches , is : @xmath64 in order to avoid the mistakes in using these formulas , below we give explicitly the definitions of all quantities which enter eqs .
( 30 - 32 ) .
let us denote the 4-momenta of the virtual photon , nucleon and resonance in the vertex @xmath65 through @xmath66 , respectively : @xmath67 the z - axis is directed along the photon 3-momentum ( @xmath68 ) in the @xmath4 rest frame , @xmath69 , and @xmath70 are the projections of the nucleon and resonance spins on the z - axis . with the definition @xmath71 , we have @xmath72 @xmath73=(0 , - @xmath61@xmath74 ) , @xmath61@xmath75 ( 34 ) and @xmath76 where in the last expression we have taken into account the gauge invariance condition : @xmath77 .
in some theoretical approaches , it is convenient to use the definition of the @xmath24 helicity amplitudes in terms of the @xmath24 form factors . for the @xmath78 resonances , we will present this definition explicitly using the form factors introduced in ref .
@xcite . the dirac equation and the dirac spinor
are : @xmath86 e is the nucleon energy . using the definitions ( 30,32,37 - 40 ) , we obtain the following relations between the @xmath24 helicity amplitudes and form factors . the @xmath87 resonances : @xmath88b,\\ & & { \cal s}_{\frac{1}{2}}=\tilde { s}b\frac{|{\bf{k}}|}{\sqrt{2}},\\ & & \tilde { s}=2(m+m)g_1(q^2)+(m - m)g_2(q^2),\nonumber\\ & & b = e\sqrt{\frac{e - m}{8mk}}.\nonumber\end{aligned}\ ] ] the @xmath89 resonances : @xmath88b,\\ & & { \cal s}_{\frac{1}{2}}=-\tilde { s}b\frac{|{\bf{k}}|}{\sqrt{2}},\\ & & \tilde { s}=2(m - m)g_1(q^2)+(m+m)g_2(q^2),\nonumber\\ & & b = e\sqrt{\frac{e+m}{8mk}}.\nonumber\end{aligned}\ ] ] for the resonances with @xmath90 , the relations between the @xmath24 helicity amplitudes and form factors can be found using the results of ref . @xcite . in ref .
@xcite , the amplitudes @xmath91 are introduced , which are proportional , respectively , to @xmath92 ; the relations between @xmath91 and the @xmath24 form factors are presented too .
these relations are quite cumbersome . for this reason , here we give only the relations between @xmath20 and @xmath93 , which for the resonances with @xmath94 , ... have the form : @xmath95 where @xmath96 .
in this section , our goal is to find the explicit relation between the @xmath0 form factors contributions to the @xmath97 multipole amplitudes and the @xmath0 vertex .
this will allow us to find the connection between @xmath22 in eqs .
( 1 ) and the sign of the ratio of the @xmath25 , @xmath23 coupling constants .
this will allow us also to check the consistency of the relative sign between longitudinal ( @xmath2 ) and transverse ( @xmath1 ) amplitudes in the definitions through multipole amplitudes ( 2 - 7 ) and hadron electromagnetic current ( 30 - 32 ) . with this aim
, we will present in detail the results of the covariant calculations of the figs .
1,2 contributions to @xmath14 for the resonances with @xmath78 . for the resonances with @xmath90 ,
the relation between @xmath22 and the sign of the ratio of the @xmath25 , @xmath23 coupling constants can be found from the results of ref .
we will present it in the end of this section along with that for the @xmath78 resonances .
we will use the definitions ( 37 - 40 ) for the @xmath0 vertices and will define the @xmath23 coupling constants according to ref .
@xcite in the following form .
the @xmath89 resonances : @xmath98 @xmath99 the @xmath87 resonances : @xmath100 @xmath101 in eqs .
( 54,56 ) , @xmath102 and @xmath103 are the energy and 4-momentum of the final nucleon in the reaction @xmath14 in the c.m.s . , @xmath104 is the pion 3-momentum in this system .
the @xmath25 coupling constant is defined according to @xmath105 where @xmath106 is the 4-momentum of the intermediate nucleon in the diagram of fig .
2(a ) . for the clarity
, we will take only that part of the nucleon electromagnetic current which is related to the @xmath107 pauli form factor : @xmath108 now let us write the matrix elements for the contributions of the fig .
1 , 2(a ) diagrams to @xmath109 : @xmath110u(p ) , ~ j^p=\frac{1}{2}^{\pm},\\ & & f = p+k = p'+q,~q^2=-k^2,\end{aligned}\ ] ] where @xmath111 , and again for the simplicity , we have taken only the part of the @xmath0 vertex related to the @xmath112 form factor . the general form of the @xmath14 matrix element according to the definition of ref .
@xcite is : @xmath113+\nonumber\\ & & + 2b_2(q^2)\left[({\cal p}\epsilon)-({\cal p } k)\frac{k \epsilon}{k^2}\right]+\nonumber\\ & & + 2b_3(q^2)\left[(q\epsilon)-(q k)\frac{k \epsilon}{k^2}\right]-\nonumber\\ & & -b_5(q^2)\left[(\gamma\epsilon)-(\gamma k)\frac{k \epsilon}{k^2}\right]+\nonumber\\ & & + b_6(q^2)(\gamma k)\left[({\cal p}\epsilon)-({\cal p } k)\frac{k \epsilon}{k^2}\right]+\nonumber\\ & & + b_8(q^2)(\gamma k)\left[(q\epsilon)-(q k)\frac{k \epsilon}{k^2}\right],\nonumber \end{aligned}\ ] ] where @xmath114 .
multipole amplitudes @xmath115 are related to the invariant amplitudes @xmath116 by @xcite : @xmath117 @xmath118 @xmath119 now from eqs .
( 59,62 ) it is easy to find the contribution of the fig .
2(a ) diagram to @xmath109 : @xmath120 these relations coincide with the commonly used expressions for the born term which corresponds to the @xmath121-channel nucleon exchange . for the contribution of the fig .
1 diagram to @xmath109 we have : @xmath122 from the relations ( 2 - 9,63 - 66,70 - 74 ) , we find for the @xmath87 resonances : @xmath123 for the @xmath89 resonances : @xmath124 now from the comparison with eqs .
( 1,46 - 49 ) and using egs .
( 54,56 ) , we find : @xmath125 using the results of ref .
@xcite , it can be shown that the same relation between @xmath22 and @xmath126 is correct also for the resonances with @xmath90 , if the @xmath23 vertices are defined in the following way : for the resonances with @xmath127 , ... @xmath128 and for the resonances with @xmath129 , ... @xmath130 where @xmath96 and @xmath131 is the generalized rarita - schwinger spinor .
in this section we will present explicit formulas for the calculation of the @xmath132 helicity amplitudes in nonrelativistic quark model .
we will present also in explicit form wave functions for the n , @xmath26 , @xmath27 , and final formulas for the @xmath133 helicity amplitudes . in nonrelativistic quark model , the matrix elements which enter eqs .
( 30 - 32 ) can be written in the form : @xmath134 where @xmath135 , @xmath136 are the radial parts of the @xmath137 wave functions , and according to eqs .
( 30 - 32,35,36 ) , we have made notations : @xmath138 , @xmath139 .
in the last part of eq .
( 82 ) , it is supposed that photon interacts with the 3-rd quark . in the calculations ,
it is convenient to define the 3-rd quark momentum in the @xmath3 and @xmath4 in the form : @xmath140 where @xmath68 is the photon 3-momentum directed along the z - axis in the @xmath4 rest frame , and @xmath141 .
the momenta , which have appropriate symmetries under @xmath142 exchanges , are : @xmath143 the matrix elements of the quark electromagnetic current have the following form : @xmath144 @xmath145 where @xmath146 is the quark mass , and @xmath147 are the quark spin wave functions in the initial and final states . the radial part of the nucleon wave function with harmonic oscillator potential is @xmath148 where @xmath149 is the harmonic - oscillator parameter . in the classification over group @xmath150 ,
the proton is the member of the octet from the multiplet @xmath151 $ ] .
the spin part of the proton wave function is : @xmath152 are @xmath153 and @xmath154 -type flavor wave functions for the octet : @xmath155 and @xmath156 are @xmath153 and @xmath154 -type spin @xmath157 wave functions : @xmath158 in the explicit form this gives : @xmath159 @xmath160 in the classification over group @xmath150 , we will consider the resonance @xmath26 as the member of the octet from the multiplet @xmath151_r$ ] .
the spin part of the @xmath26 wave function in this case is the same as for the nucleon ( 91,98,99 ) , and the radial part is : @xmath162 the @xmath161 helicity amplitudes obtained using the definitions ( 30,32 ) and the results presented in this section are : @xmath163 in the classification over group @xmath150 , we will consider the resonance @xmath27 as the @xmath165 member of the multiplet @xmath166 $ ] .
the spin part of the @xmath27 wave function in this case is : @xmath167 where @xmath168 the radial part of the @xmath27 wave function is : @xmath169 the @xmath164 helicity amplitudes obtained according to the definitions ( 30,32 ) and the results of this section are : @xmath170
in the framework of quark model , the signs of the @xmath15 coupling constants were found in ref .
@xcite for the resonances of the multiplet @xmath166 $ ] using an approach based on pcac .
there are also results on the sign of the @xmath171 coupling constant obtained in refs .
@xcite , respectively in @xmath172 model and using pcac ; they coincide with each other .
quark model predictions for the @xmath132 helicity amplitudes are presented in refs .
@xcite taking into account the @xmath15 signs obtained in these papers .
however , traditionally , quark model predictions for the @xmath132 helicity amplitudes are presented with the common sign fixed by taking the sign of @xmath173 at @xmath174 equal to that of the amplitude extracted from experimental data .
sometimes , such definition of the sign can bring to confusing and wrong results , as we will demonstrate below on the example of the @xmath26 resonance . transition .
the full circles are the data extracted from the jlab - clas @xmath10 data @xcite .
the bands present model uncertainties of the data .
the full triangle at @xmath174 is the rpp estimate @xcite .
the red curves correspond to the predictions of the light - front relativistic quark models : dashed - ref .
@xcite , solid - ref .
@xcite , dotted - ref .
@xcite , dashed - dotted - ref .
the blue curves are the results obtained via nonrelativistic calculations : solid - ref .
@xcite , dashed - ref .
@xcite . the plots ( c ,
d ) present the predictions in that form as they are given in the papers . in plots ( a , b )
, all results are presented with correct signs .
, width=359 ] in fig .
3 , we present the results for the @xmath161 helicity amplitudes extracted from the jlab - clas data on the @xmath10 reactions @xcite in comparison with the quark model predictions obtained in the light - front dynamics @xcite and via nonrelativistic calculations @xcite . for clear understanding of the results presented in fig .
3 , it is important to note that the @xmath173 amplitude found with the same @xmath3 and @xmath26 wave functions via nonrelativistic calculations and in the light - front relativistic approaches have at @xmath174 opposite signs . for the first time
this was mentioned in ref . .
however , in the presentation of the results in traditional way ( see plots ( c , d ) ) , the amplitudes @xmath173 at @xmath174 from refs . @xcite and @xcite appear with the same sign . as a result ,
for higher @xmath7 , we have strong disagreement between the corresponding predictions , and the quark model predictions @xcite strongly disagree with the data extracted from experiment . in plots ( a , b ) , we present all results with the @xmath171 sign found in refs .
we have also corrected the relative signs between the @xmath173 and @xmath2 amplitudes from refs .
it can be seen that for @xmath175 , this results in good agreement with experiment for the signs of amplitudes from all approaches @xcite ; we have also better agreement of the results obtained in different approaches with each other . for the resonances from the multiplet
@xmath166 $ ] : n(1520)d@xmath9 , n(1535)s@xmath8 , @xmath176s@xmath177 , n(1650)s@xmath8 , n(1700)d@xmath9 , and @xmath178d@xmath179 , the signs found in traditional way and in ref
. @xcite coincide with each other for all resonances , except n(1700)d@xmath9 . with the @xmath15 coupling constants found in ref .
@xcite , nonrelativistic quark model prediction for the @xmath180 amplitude at @xmath174 is positive for the resonances n(1535)s@xmath8 , @xmath176s@xmath177 , n(1650)s@xmath8 , n(1700)d@xmath9 , and @xmath178d@xmath179 , and negative for n(1520)d@xmath9 .
by performing covariant calculations of the resonance ( fig . 1 ) and born terms ( fig .
2 ) contributions to @xmath181 , we have found the relations between the following definitions of the @xmath17 helicity amplitudes @xmath182 : \(i ) the definition through the @xmath181 multipole amplitudes which is commonly used for the extraction of @xmath182 from experimental data on @xmath181 ; \(ii ) the definitions through the hadron electromagnetic current and the @xmath183 form factors which are used in theoretical calculations .
these relations include the relative sign between the coupling constants @xmath184 and @xmath185 which we have defined explicitly for the vertices @xmath23 and @xmath25 . for the proper comparison of theoretical predictions with the results extracted from experimental data , it is important to have both quantities , the @xmath17 amplitudes and the sign of the ratio @xmath186 , calculated within the same theoretical approach
. this is demonstrated on the example of the @xmath161 transition .
the performed calculations allowed us also to check and present the formulas for the @xmath17 helicity amplitudes through the hadron electromagnetic current and the @xmath183 form factors which give the correct relative sign between the longitudinal @xmath2 and transverse @xmath1 amplitudes , consistent with that for the amplitudes extracted from experiment . to avoid sources of mistakes in the calculation of this sign , we give definitions of all quantities which enter the formulas . for completeness , we have presented formulas for the calculations of the @xmath187 helicity amplitudes in nonrelativistic quark model along with the n , @xmath26 , @xmath27 wave functions and final formulas for the @xmath133 helicity amplitudes . taking into account the available information on the signs of the ratios @xmath186 in the framework of quark model , we have presented , in the case of nonrelativistic quark model , the signs of the @xmath173 amplitudes for the resonances from the multiplet @xmath166 $ ]
. this may be useful for the correct presentations of the signs of the helicity amplitudes @xmath182 in quark model .
i. g. aznauryan is grateful to ya .
azimov and j. l. goity for useful discussions .
this work was supported in part by the u.s .
department of energy and the national science foundation .
jefferson science associates , llc , operates jefferson lab under u.s .
doe contract de - ac05 - 060r23177 . | we present definitions and formulas that may be useful for a consistent computation of helicity amplitudes for the process @xmath0 in theoretical approaches . of particular importance
is the correct determination of the common sign of the amplitudes and of the relative sign between the transverse ( @xmath1 ) and longitudinal ( @xmath2 ) amplitudes .
this clarification is necessary to clear up confusions present in theoretical works .
using the definitions presented in this paper will enable a direct comparison with amplitudes extracted from experimental data . |
since the suggestion of @xmath0 suppression as a signal of quark - gluon plasma ( qgp ) formation by matsui and satz @xcite in 1986 the problem of quarkonium dissociation in hot and dense strongly interacting matter has played a key role for qgp diagnostics in relativistic heavy - ion collision experiments .
the original idea was that in a qgp the string tension of the confining potential vanishes and the residual one - gluon exchange interaction undergoes a debye screening by the color charges of the plasma .
when the temperature dependent debye radius @xmath1 ( the inverse of the debye mass @xmath2 ) becomes shorter than the bohr radius of the charmonium ground state ( @xmath0 ) then the mott effect @xcite ( bound state dissociation ) occurs and the corresponding temperature is @xmath3 .
this simple idea grew up to a multifacetted research direction when not only in the first light ion - nucleus collisions at the cern na38 experiment , but also in proton - nucleus collisions at fermilab @xmath0 suppression has been found so that there is not only a qgp but also a cold nuclear matter effect on charmonium production , see @xcite for a recent review . if one wants to explore the question of screening in a plasma more in detail then a variety of appraoches is available in the literature , from the original debye - hckel approach @xcite where one can study any vacuum potential ( for example the cornell potential ) and see its medium modification , over the thermodynamic green functions approach to the ab - initio studies of heavy - quark potentials in lattice qcd . with the obtained medium - dependent potentials one can then study the bound state problem by solving the nonrelativistic schrdinger equation or , more systematically
, the thermodynamic @xmath4 - matrix for quarkonia @xcite . on the other hand one may calculate proper correlators directly from lattice qcd and extract from them spectral functions @xcite .
there is an intriguing disagreement between the mott temperatures deduced from these spectral functions and those of the potential models : the latter are much smaller than the former ! from the lattice data for quarkonium correlators one
has extracted @xmath5 while in potential model calculations @xmath6 .
this problem has lead to the discussion of the proper thermodynamical function to be used as a potential in the schrdinger equation .
should it be the free energy or the internal energy ?
we will not follow this question in the present work , but refer to @xcite and references therein . here
we examine a simple model of screening as derived from one - loop calculations in thermal quantum field theory , and make a small extension of this result by putting the internal fermion lines in a constant temporal gauge field which mimics confining gluon dynamics ( polyakov - loop potential ) . in our approach the medium is made of plasma of massless quarks , described by the chirally symmetric phase of the nambu jona - lasinio ( njl ) model of qcd @xcite .
confinement is implemented in the most simple way by coupling the system to the poyakov loop variable - resulting in the so called polakov loop njl - model ( or pnjl model ) .
recently it has been successfully used to reproduce lattice data @xcite or to describe light meson physics at finite temperatures and densities @xcite .
given the self energy ( polarization function ) of a boson that mediates the interaction , the screened potential is given by a resummation of one - particle irreducible diagrams ( `` bubble '' resummation = rpa ) @xcite @xmath7~ , \label{vsc}\ ] ] where we take @xmath8 , @xmath9 as the unscreened vacuum potential .
the longitudinal gluon polarization function @xmath10 in the finite @xmath11 case can be calculated within perturbative thermal field theory where it takes the form @xmath12~,\ ] ] where @xmath13 are the bosonic and @xmath14 are the fermionic matsubara frequencies of the imaginary - time formalism .
the symbol @xmath15 stands for traces in color , flavor and dirac spaces .
@xmath16 is the propagator of a massless fermion coupled to the homogeneous static gluon background field @xmath17 .
its inverse is given by @xcite @xmath18 where @xmath17 is related to the polyakov loop variable defined by @xcite @xmath19 the physics of @xmath20 is governed by the temperature - dependent polyakov loop potential @xmath21 , which is fitted to describe the lattice data for the pressure of the pure glue system @xcite . after performing the color- , flavor- and dirac traces and making the fermionic matsubara summation , we obtain in the static , long wavelength limit @xmath22 where @xmath23 is the debye mass , the number of degrees of freedom is @xmath24 and @xmath25 is quark distribution function @xcite .
the screened potential is thus @xmath26~.\ ] ] in comparison to the free fermion case @xcite the coupling to the polyakov loop variable @xmath20 gives rise to a modification of the debye mass , given by the integral @xmath27 in the limit of deconfinement ( @xmath28 ) , the case of a massless quark gas is obtained ( @xmath29 ) while for confinement ( @xmath30 ) one finds that @xmath31 .
for the temperature dependence of @xmath20 we employ in the following chapter the results of a nonlocal pnjl model @xcite .
here we will use the derived potential in the quantum mechanical way to estimate the dissociation temperature . with the trial wave function for the 1s state @xmath32 and the non - relativistic two - body hamiltonian @xmath33 where the potential term is the fourier transform of screened coulomb potential ( [ vs ] )
, we obtain the energy functional for the ritz variational principle @xmath34 simultaneously satisfying the conditions for the ground state energy @xmath35 and for a vanishing binding energy ( mott effect ) , @xmath36 , provides us with an analytic expression for the critical debye mass @xmath37 using once again the condition @xmath38 , results in the mott condition for the debye potential @xcite @xmath39 where @xmath40 is the bohr radius and @xmath41 the binding energy of ground state in the vacuum ( @xmath42 ) . due to the temperature dependence of the debye mass
, we obtain the mott dissociation temperature in the massless quark gas ( for @xmath43 ) @xmath44 [ cols="^,^,^",options="header " , ] [ tab : limits ] in tab .
[ tab : limits ] we give the parameter values according to set ( i ) of ref .
the poyakov loop variable contribution considered in previous section affects the mott temperature in eq .
( [ mottcondition ] ) and gives the following formula @xmath45 this means that in the case of confining color correlations ( @xmath46 ) the debye screening radius is larger than in a free quark gas at the same temperature , so that bound states get stabilized against thermal dissociation by color screening . for charmonium ( upper panel ) and bottomonium ( middle panel ) with ( solid lines ) and without ( dashed lines ) coupling to the polyakov loop @xmath20 shown in the lower panel ( from @xcite ) .
sequential dissociation of quarkonia states occurs at the mott temperatures @xmath47 for which their bohr radius equals @xmath1.,scaledwidth=65.0%,scaledwidth=60.0% ] the influence of the poyakov loop variable on the dissociation temperature is summarized in figure [ fig : debye ] which shows the temperature dependence of the debye radius @xmath48 compared to the bohr radii of the low - lying states of the charmonium and bottomonium family , respectively . due to its larger mass and smaller bohr radius , the @xmath49 dissociates at higher temperatures than the @xmath0 , where the free quark gas case is almost reached and the mott temperatures which almost coincide for the cases with and without coupling to the polyakov loop field . for the lighter @xmath0
there is a noticeable stabilization due to the coupling to the polyakov loop potential which results in an increase of the mott dissociation temperature from @xmath50 to @xmath51 , still more similar to results of nonrelativistic potential models rather than the still higher dissociation temperatures conjectured from the spectral functions deduced from lattice data for heavy quarkonium correlators by the maximum entropy method . for estimating the mott temperatures of the excited quarkonia states we have employed the scaling of bound state radii : @xmath52 , @xmath53 as obtained in the cornell - type potential model @xcite .
we have applied the methods of thermal field theory to estimate the effects of debye screening on heavy quarkonia bound state formation . in order to account for residual effects of confining color correlations in the deconfined phase ,
we have used the pnjl model in the evaluation of the one - loop polarization function . as expected ,
a stabilization of bound states in the vicinity of the critical temperature for @xmath54 is obtained .
we applied ritz variational principle to derive the mott criterion for bound states of the statically screened debye potential and obtained mott temperatures in good agreement with previous results from nonrelativistic potential models exploiting lattice qcd singlet free energies as potentials in the schrdinger equation for heavy quarkonia .
this agreement with previous results ( see , e.g. , ref .
@xcite ) includes also higher quarkonia resonances for which the stabilization effects is more pronounced .
jj received financial support from the bogoliubov - infeld program for his participation in the helmholtz international summer schools in dubna .
the work of db has been supported in part by the polish ministry for science and higher education under grant no .
n n 202 0953 33 and by rfbr grant no .
08 - 02 - 01003-a .
m. asakawa , t. hatsuda and y. nakahara , prog .
phys . * 46 * 459 ( 2001 ) ; + a. jakovac , p. petreczky , k. petrov and a. velytsky , phys . rev .
d * 75 * 014506 ( 2007 ) .
h. satz , j. phys .
g * 32 * r25 ( 2006 ) .
a. beraudo , j. p. blaizot and c. ratti , nucl .
a * 806 * 312 ( 2008 ) .
d. blaschke , m. buballa , a. e. radzhabov and m. k. volkov , yad .
fiz . * 71 * , 2012 ( 2008 ) .
f. arleo , p. b. gossiaux , t. gousset and j. aichelin , phys .
d * 65 * 014005 ( 2002 ) . | we investigate the mott effect for heavy quarkonia due to debye screening of the heavy quark potential in a plasma of massless quarks and antiquarks .
the influence of residual color correlation is investigated by coupling the light quark sector to a temporal gauge field driven by the polyakov loop potential .
this leads to an increase of the mott dissociation temperatures for quarkonia states which stabilizes in particular the excited states , but has marginal effect on the ground states . |
the equation of state ( eos ) of neutron star matter is intimately related to the physics of the strong interactions between fundamental particles and therefore it is of great relevance in high energy and particle physics .
theories on the eos of neutron
star matter at supra
nuclear densities provide a firm upper limit on the mass for each eos .
the stiffer the eos , the higher the mass limit . neutron stars with masses well above @xmath7 can not exist for so
called soft equations of state , in which matter at high densities is relatively compressible ( e.g. , due to meson condensation or a transition between the hadron and quark gluon phases ; cf .
therefore , measuring a high mass for even one neutron star would imply the firm rejection of many proposed soft eoss ( e.g. see discussion by @xcite ) .
neutron star masses of more than @xmath8 are excluded assuming ( among other things ) that general relativity is the correct theory of gravity and that the velocity of sound is less than the velocity of light ( @xcite ) . therefore , a lower limit on the mass of the compact object of 3 @xmath9 or more implies a black hole compact object .
the most accurately measured neutron star masses we have are from precise radio timing measurements of double neutron
star binaries .
these empirical masses cluster around @xmath10 ( @xcite ) .
this is close to the mass of 1.32 @xmath9 predicted from theoretical model calculations of type ib supernovae ( @xcite ) .
however , these double neutron star binaries are formed in high mass x ray binaries and due to the short life of the massive companion only a limited amount of matter can be accreted .
in contrast , in low mass x
ray binaries ( lmxbs ) , the rapid rotation and low magnetic field of the neutron stars are considered evidence for much larger amounts of accreted matter ( see @xcite for a review ) . indeed , from binary evolution models , lmxbs could accrete up to 0.7@xmath11 ( e.g. ) .
if so , the masses of their neutron stars could be @xmath12 .
most of the constraints from timing observations of the successors of these lmxb primaries , the millisecond radio pulsars , are weak ( @xcite ) , as the almost purely circular orbits of these systems often prevents one from measuring most of the post newtonian parameters that underlie the precise masses for the double neutron star binaries . with masses of @xmath13 and 2.2@xmath14 ,
the msec pulsars psr b1855@xmath1509 and psr j0751 + 1807 are exceptions to this ( @xcite ; @xcite , 95% conf . ) . dynamical mass estimates or dynamically determined limits on the mass of the neutron star in lmxbs are available for the two transient sources cen x4 and xte j2123058 , the z source cyg x2 , and the pulsar 2s 1822371 ( cen x4 ; @xmath160.6 @xmath11 @xcite ; xte j2123058 ; @xmath170.31 @xmath11 @xcite ; cyg x2 ; @xmath180.23 @xmath11 @xcite ; @xcite ; 2s 1822371 ; @xmath190.06 @xmath11 ; @xcite ; @xcite ) .
the problem is that many of the companion stars are small , late type stars which at a distance of more than a few kpc are too faint to observe , especially since the integration times are limited to typically 1/20th of the orbital period in order to avoid doppler smearing of the spectral lines due to the binary motion .
furthermore , the binary inclination is often poorly constrained and effects of irradiation shift the centre - of - light with respect to the centre - of - mass
. some of the aforementioned problems will be less severe when observing long orbital period systems for which the inclination is well constrained from the fact that eclipses are observed : 2s 0921630/v395 car is such a high inclination lmxb with a long orbital period ( p@xmath20 days ; e.g. @xcite , @xcite , and @xcite ) .
the inclination of 2s 0921630 is relatively well constrained and must be high since partial eclipses of the compact object and accretion disc have been observed in both x rays and optical wave bands ( e.g. @xcite ; ; @xcite ) .
2s 0921630 is thought to be a halo object in orbit with a k01 iii companion star ; absorption lines of the companion have been detected ( @xcite ; ) .
the corresponding radial velocity curve would provide a limit to the mass of the companion star .
in this paper we report on spectroscopic observations of 2s 0921630/v395 car .
we observed 2s 0921630/v395 car with the fors2 spectrograph mounted on the yepun very large telescope ( vlt ) . in the period dec 26 , 2003
mar 13 , 2004 , 22 spectra using the 1200r+93 and 22 spectra using the 1028z+29 grating have been obtained in service mode .
the exposure time was 1300 s for each observation and a slit
width of 0.4 `` was used on each occasion .
the dispersion was 0.75 per pixel for the spectra obtained with the 1200r+93 grating and 0.86 per pixel for the spectra obtained with the 1028z+29 grating . with a slit
width of 0.4 '' the two pixel resolution is approximately 65 km s@xmath1 at 6500 . after each observing night helium neon argon lamp wavelength calibration spectra were obtained .
the seeing varied between 0.5 and 1.4 and the spectra were obtained when the source had an airmass of @xmath21 , except on one occasion on jan 20 , 2004 when the airmass was nearly 2 .
the spectra were bias subtracted , flatfield corrected , optimally extracted and wavelength calibrated using the iraf reduction package .
the rms scatter in the wavelength calibration was @xmath220.03 .
we further applied a small correction to the wavelength solution by aligning sky lines observed in each of the spectra . due to variations in seeing , resulting in variable slit losses , the signal to noise ratio varied between the spectra from @xmath2280240 over the wavelength range 59206520 .
next , we exported the extracted spectra to the data analysis package molly . in molly the observation times
were first corrected to the heliocentric julian date time frame ( using utc times ) .
next , we normalised and rebinned the spectra to a uniform velocity scale removing the earth s velocity .
below , we first discuss results obtained using the 1200r+93 grating spectra .
we cross correlated the 1200r+93 grating spectra with spectra of template stars rebinned to the same velocity scale .
nine template star spectra with spectral types ranging from g5k7 had been obtained with the keck telescope at similar resolution .
we fitted a sinusoid to the measured velocities as a function of time .
the fit - parameters are : the phase , the semi amplitude , and the period of the sinusoid and the systemic velocity . however , since the errors on the velocity represent the statistical error bars only , they do not include the systematic effects which we found to be important ( we will come back to this in the discussion ) .
therefore , the reduced @xmath23 of the fit was much larger than 1 . in order to estimate realistic errors on the fit parameters , we increased the size of the error bars such that the reduced @xmath23 was close to 1 .
we found a best fit period of 9.006@xmath24 days , consistent with the orbital period of @xmath229 days found before ( here and below we give 1 @xmath3 error bars ) . from this
we derive an orbital ephemeris of hjd = 2453000.49(8 ) + n@xmath25 ( utc ) where phase zero is defined as superior conjunction of the neutron star and the number in between brackets denotes the uncertainty in the last digit .
the derived ephemeris does not significantly depend on the spectral type of the used template star and is consistent within the errors with the ephemeris given by @xcite .
we determined a radial velocity semi
amplitude of [email protected] km s@xmath1 using the k1v template star hd 124106 ( see figure [ radvel ] ) . the radial velocity semi - amplitude does not significantly depend on the spectral type of the template star . i.e. the amplitudes we found varied by 2 km s@xmath1 between the minimum and maximum amplitude for template spectral types in the range of g5k7 .
we found a systemic velocity of [email protected] km s@xmath1 . after correcting each individual spectrum for the observed sinusoidal velocity shift appropriate for the orbital phase we created an average spectrum in the frame of the donor star for the data of 2s 0921630 .
one can in principle derive the rotational broadening from an optimal subtraction ( see for the description of this technique @xcite ) .
however , our data is not well suited for this since the dispersion at which both the template star and the source were observed is relatively low ( @xmath26 km s@xmath1 per pixel in the 59206520 spectral range for the template star , this spectrum was resampled to the dispersion of @xmath27 km s@xmath1 per pixel in the 59206520 spectral range at which the source was observed ) . since the template stars were not obtained with the same telescope and instrumental set up , the instrumental profile will be different from that of the object spectra ; together with the low resolution this means that it is not possible to reliably measure the rotational broadening when it is of the order of the spectral resolution .
earlier results on the rotational broadening of 64@xmath09 km s@xmath1 measured by showed that this is likely the case . found a best fit spectral type of k0 for the companion star ; we found that the residuals were smallest when we used the k1 template star hd 124106 ( irrespective of broadening ) .
we further found that on average 20 per cent of the light comes from the companion star in the 59206520 range . in figure [ optsub ]
( _ left panel _ )
we plot the average spectrum in this range , the broadened template star veiled by a source of constant light contributing 80 per cent of the light , and the residuals after subtraction the veiled template star spectrum .
we used the spectral range from 59206520 for the optimal subtraction , however , the range in wavelengths between 62506350 was masked since it contains a strong interstellar absorption feature .
we further examined whether the contribution of the k1 companion star to the total amount of light in the 59206520 range varies as a function of the binary orbital phase .
we averaged spectra obtained at the same phase using phase bins of width 0.1 .
we optimally subtracted the template spectrum of the k1 star from the average spectra in each phase bin .
the measured fractional contribution of the companion star to the total amount of light in the 59206520 range is plotted in figure [ compfrac ] . the large difference in fractional contribution from the k
star between phase 0 and phase 0.1 is difficult to explain . besides the stellar absorption lines apparent in the wavelength range 59206520 shown in the top spectrum of the _ bottom left panel _ in figure [ optsub ] , additional stellar absorption lines were present in the range 65206800 (see the _ top left panel _ in figure [ optsub ] ) .
furthermore , several emission lines are present in the spectrum .
unfortunately , our observations lack the resolution to resolve these accretion disc lines and hence only one broad emission peak was observed whereas e.g. observed that the h@xmath28 and he i emission lines at 6562.76 and 6678.15 , respectively have the double peaked profile typically observed in accretion disc spectra .
we detected a broad emission peak at @xmath226820 which we could not positively identify .
we checked whether ccd defects were present at the position relevant for this wavelength in the flatfield images or in the individual images but none were found .
so far , we have only discussed spectra obtained with the 1200r+93 grating . in figure [ optsub ] ( _ bottom right panel _ ) we show the phase folded average spectra of the 1028z+29 grating in the range between 83008900 .
clear donor star features such as the ca ii triplet at 8498 , 8542 , and 8662 , and the mg i line at 8806.75 can be seen at orbital phases where the back , non irradiated side , of the companion star is observed .
furthermore , superimposed on the paschen disc emission lines , paschen absorption lines can be seen .
unfortunately , the combined effect of two unresolved broad accretion disc emission lines , a paschen absorption line and , at some wavelengths , and at certain orbital phases , a ca ii absorption line is too difficult to disentangle with the limited resolution data in hand .
therefore , we did not use these spectra for our radial velocity study .
besides the features visible in the _ bottom right panel _ of figure [ optsub ] we identified two other lines in the 1028z+29 grating spectra ; a weak fe i absorption line complex near 7750 and a paschen emission line at 9229 (see the _ top right panel _ in figure [ optsub ] ) .
we determined the radial velocity curve of the companion star in 2s 0921630/v395 car using vlt / fors2 spectroscopic observations .
a sinusoidal fit to the radial velocity measurements with a semi amplitude , k@xmath29 , of [email protected] km s@xmath1 and with a systemic velocity , @[email protected] km s@xmath1 represents the data well . in these error estimates
we artificially increased the error bars on the individual velocity measurements by a factor 4 in order to obtain a reduced @xmath23 of the fit close to 1 . before we discuss the implications of the k velocity , we investigate possible reasons for the large reduced @xmath23 obtained when using the formal statistical errors only . near quadrature at phase 0.25
the velocity measurements obtained on different nights differ well beyond the statistical errors . from figure [ compfrac ]
we can see that the companion star contribution to the total amount of light in the 59206520 range varies as a function of orbital phase .
the phasing of this variation suggests that the centre of light emitted by the k
star is likely to be shifted with respect to its centre of mass .
an explanation is that the inner side of the companion star is heated by the ( x ray ) irradiation coming from ( near ) the compact object , reducing the equivalent width of the k
star stellar lines in the range 59206520 ( see also e.g. @xcite ) .
this manifests itself in several ways : a lower fractional contribution to the total amount of light from the k star at e.g. phase 0.5 when using the optimal subtraction technique ( e.g. figure [ compfrac ] ) an increase in velocity ( especially near quadrature ) with respect to the velocity associated with the centre of mass of the companion star , since the line profiles will , as a result of the stellar rotation , be preferentially shifted to either a larger red- ( phase 0.25 ) or blue - shift ( phase 0.75 ) .
hence , variations in the amount of ( x ray ) irradiation between the observations could explain the variations in velocities near quadrature as observed in figure [ radvel ] .
since the observations were obtained with at least one day in between , with a baseline of several months , it seems likely that the irradiation changed between the observations .
furthermore , the velocity measurements around phase 0.5 do not fall on the best
fit sinusoid .
those at phases 0.40.5 lay above the best
fit sinusoid , whereas those at phases 0.50.6 fall below the curve .
this could well be explained by the rossiter effect ( cf .
such deviations occur if the companion star is partially eclipsed by the accretion disc .
such a partial eclipse of the companion star can also help explain the low fraction of light coming from the companion star at phase 0.5 ( see figure [ compfrac ] ) .
the radial velocity measurements at phases just prior to phase 0.5 , the partial eclipse of the companion star , are biased to higher , red shifted velocities since the part of the companion star that is eclipsed at those phases is mainly rotating towards the observer .
hence , the rotational broadening of the line will be biased towards the red .
the reverse holds for phases just after the partial eclipse . for the rotational
broadening of the stellar lines we take @xmath31 km s@xmath1 as found by . for a roche lobe filling companion star
the following relation holds : @xmath32^\frac{1}{3}$ ] ( e.g. @xcite ; q is defined here as m@xmath29 divided by the mass of the compact object , m@xmath33 ) .
before we can determine q , and hence with k@xmath29 the mass of the compact object , we have to take effects of the non uniform absorption distribution on the stellar surface caused by irradiation into account . to this end
, @xcite developed a so called k correction .
following the procedure laid out in @xcite , @xmath34 .
@xmath35 is the corrected observed ( @xmath36 ) radial velocity semi amplitude and @xmath37 . here
@xmath38 is the distance between the centre - of - mass of the binary and the centre - of - mass of the companion star and @xmath39 is a geometrical correction factor less than 1 . in the extreme case that the hemisphere facing the compact object does not contribute to the observed stellar absorption lines at all and the other hemisphere has a uniform absorption line strength , @xmath40 ( @xcite ) . since we still have a contribution of the k1 star of around 7 per cent at phase 0.5 we know that this is an overestimation of the k correction in the case of 2s 0921630 .
hence , applying this upper limit on the k
correction , @xmath41 km s@xmath1 , we get a firm lower limit of k@xmath42 km s@xmath1 . plugging in the numbers given above yields [email protected]@xmath44
; note that in deriving this we used the equation for @xmath45 of @xcite which is valid for @xmath46 .
however , the difference with the equation of @xcite is small for [email protected] and does not change the result significantly given the large uncertainty in @xmath48 .
the uncertainty in q is dominated by the large error on the rotational velocity .
thus , we find for the minimum mass of the compact object , @xmath49 . as mentioned in the introduction , the inclination of 2s 0921630 must be high .
furthermore , we have presented evidence for a partial eclipse of the companion star . for inclinations ,
@xmath50 ( cf .
@xcite ; @xcite ) the mass of the compact object is @xmath51 .
using @xmath52 , we get for the radius of the companion star , @xmath53 , @xmath54 . under the assumption that the companion star fills its roche lobe
, the mass of the donor star , @xmath55 , depends only on its radius and the orbital period ( cf .
@xcite ) , which gives @xmath56 ( here the orbital period is measured in hours , and the mass and radius of the companion star are in solar units ) .
this gives @xmath57 .
this m@xmath29 suggests an interesting evolutionary stage of the binary .
m@xmath29 is more massive than standard donor stars in lmxbs .
recently , there have been a number of evolutionary calculations ( e.g. @xcite ; @xcite ) showing that systems with donor masses as high as @xmath585 m@xmath59 can enter a relatively long phase of mass transfer to a neutron star after the donor has lost most of its mass .
2s 0921630 would nicely fit in this scenario , evolving along the track of an initial 34 m@xmath59 donor .
however , in this evolutionary scenario the current time - averaged mass - transfer rate is expected to be high , certainly super - eddington .
this is at odds with the finding from studying x - ray lines as measured with _
chandra _ and xmm_newton _ that the observed x - ray luminosity of @xmath60 erg s@xmath1 ( for a distance of @xmath587 kpc ) is likely close to the intrinsic source luminosity ( @xcite ) .
perhaps the current mass - transfer rate is much lower than the time - averaged transfer rate , or the donor mass and/or the mass - ratio are overestimated .
above , we assumed that the companion star is in co rotation with the binary orbit .
synchronization of the rotational period and the binary period occurs on timescales shorter than the circularization timescale ( e.g. @xcite ) and no signs for orbital eccentricity have been found ( see figure [ radvel ] ) .
furthermore , using the companion star parameters derived above and equation 7 of @xcite we find that the synchronization timescale is @xmath61100 yr for 2s 0921630 , justifying this assumption .
we conclude that the compact object in 2s 0921630 is likely to be a massive neutron star or a low - mass black hole . if indeed the compact object in 2s 0921630 is a massive neutron star the mass measurement would rule out soft equations of state .
future high resolution optical observations should provide a much more accurate handle on the rotational velocity of the companion star , yielding a more accurate constraint on the mass of the compact object .
support for this work was provided by nasa through chandra postdoctoral fellowship grant number pf340027 awarded by the chandra x ray center , which is operated by the smithsonian astrophysical observatory for nasa under contract nas839073 .
ds acknowledges a smithsonian astrophysical observatory clay fellowship .
gn is supported by pparc . the use of the spectral analysis software package molly written by tom marsh is acknowledged . | in this paper we report on optical spectroscopic observations of the low - mass x - ray binary 2s 0921630 obtained with the very large telescope .
we found sinusoidal radial velocity variations of the companion star with a semi amplitude of [email protected] km s@xmath1 modulated on a period of [email protected] days , consistent with the orbital period found before for this source , and a systemic velocity of [email protected] km s@xmath1 . due to
x ray irradiation the centre of light measured by the absorption lines from the companion star is likely shifted with respect to the centre of mass .
we try to correct for this using the so called k correction . conservatively applying the maximum correction possible and using the previously measured rotational velocity of the companion star
, we find a lower limit to the mass of the compact object in 2s 0921630 of @xmath2 ( 1 @xmath3 errors ) .
the inclination in this system is well constrained since partial eclipses have been observed in x - rays and optical bands . for inclinations between 60@xmath4
we find @xmath5 .
however , using this maximum k correction we find that the ratio between the mass of the companion star and that of the compact object , q , is @xmath6 implying super eddington mass transfer rates ; however , evidence for that has not been found in 2s 0921630 .
we conclude that the compact object in 2s 0921630 is either a ( massive ) neutron star or a low mass black hole .
stars : individual ( 2s 0921630 ) accretion : accretion discs stars : binaries stars : neutron x - rays : binaries |
much attention has been devoted in the past years to the study of heat conduction in low - dimensional systems @xcite .
such study is not only important for understanding the fundamentals of statistical mechanics , but also for potential applications in heat control and management in nanoscale devices .
indeed , various models for thermal rectifiers / diodes that allow heat to flow easily in one direction have been proposed @xcite . moreover ,
based on the phenomenon of the _ negative differential thermal resistance _ ( @xmath0 ) observed in nonlinear lattices , a theoretical model for thermal transistor " is also proposed @xcite , which allows us to control heat current ( due to phonons ) by adjusting the temperature of the gate terminal ( or called gate temperature " ) .
this is similar to the electronic transistor which controls the electric current by adjusting the gate voltage .
we should point out that our thermal transistor " is different from the recent heat transistor " , which controls the heat flux ( caused by electrons ) via adjusting the gate voltage @xcite .
most recently , based on the thermal transistor , the different thermal logic gats have been also realized theoretically@xcite , which means that the phonons , the heat carrier , can also be used to carry information and processed accordingly .
the purpose of the current paper are two folds .
first , we shall discuss how to improve significantly the switching efficiency of 1d thermal transistor ; second , we shall extend the study from 1d model to 2d which is closer to laboratory fabricated materials .
the 1d thermal transistor model consists of three segments @xmath1 , @xmath2 , and @xmath3 and each segment is a frenkel - kontorova ( fk ) lattice @xcite .
the model configuration , which is shown in fig .
1(a ) , is similar to the one proposed in ref.@xcite except for an additional interface coupling @xmath4 . in each segment , we regard the particle that is coupled to heat bath as the first particle and the interface particle as the last particle .
the total hamiltonian of the 1d model writes @xmath5and the hamiltonian of each segment can be written as @xmath6=@xmath7 , with @xmath8 and @xmath9 denote the displacement from equilibrium position and the conjugate momentum of the @xmath10 particle in segment @xmath11 , where @xmath11 stands for @xmath1 , @xmath2 or @xmath3 .
the parameters @xmath12 and @xmath13 are the harmonic spring constant and the on - site potential of the fk lattice .
we couple the last particle of segment @xmath1 , @xmath2 , and @xmath3 to particle @xmath14 via harmonic springs .
thus @xmath15=@xmath16 .
fixed boundaries are used , i.e. , @xmath17=0 and each segment consists of 10 particles .
= 0.01 , @xmath18=5.00 , @xmath19=0.35 , @xmath4=0.30 ; @xmath20=0.20 , @xmath21=1.00 , @xmath22=0.20 , @xmath23=0.05 ; @xmath24=5.00 , @xmath25=1.00 , @xmath26=0.20 , and @xmath27=5.00 .
( c ) thermal modulator : all parameters remain the same as in ( b ) except with @xmath26=0.05 . ] in our simulations , we use the nos - hoover heat baths @xcite and we set the mass of each particle and the boltzmann constant to unity unless otherwise stated . in the nonequilibrium stationary state , the heat flux , which is the rate of energy transport , is constant along each segment and the local equilibrium temperature is given by the time average of the square of particle s velocity per degree of freedom .
we show in fig .
1 that the model described by hamiltonian ( 1 ) can display different heat controlling features depends on the applied gate temperature , @xmath28 .
in particular , the model can act as a heat switch , namely , the system can be switched from a heat insulator to a heat conductor by just adjusting the gate temperature . as demonstrated in fig .
1(b ) , when @xmath28 is at 0.01 , 0.08 , and 0.14 , @xmath29@xmath30@xmath31 , thus @xmath32=0 , namely , the control terminal does not provide any current to the system , the system , however is switched from an insulating ( off ) state , to a semi - conducting ( semi - on ) state , and conducting ( on ) state . at these three points ,
@xmath33 are 3.8x@xmath34 , 1.8x@xmath35 , and 3.2x@xmath35 , respectively .
the ratio of @xmath36 at the on " state to that at the off " state is considered as the switching efficiency which is about 800 .
we further demonstrate that the model can act as a heat modulator as well in which @xmath36 can be continuously modulated by the gate temperature . to show this
, we recall that for any two weakly coupled lattices , the heat current depends among others , on the strength of the interface coupling used @xcite .
thus , to have @xmath37@xmath38@xmath39 , we can reduce the strength of @xmath26 so that the heat flow through the control terminal decreases . apparently , @xmath26 should not be too small otherwise @xmath40 is not well - controlled by @xmath28 .
the numerical result is shown in fig .
1(c ) . in our next configuration
, we replace segment _
d _ with a fpu-@xmath41 lattice whose phonon spectrum is contributed by low frequency vibrations ( acoustic phonons ) only .
the hamiltonian is given by @xmath42=@xmath43 , with @xmath41 being the anharmonicity .
we show in fig .
2(a ) that such a setup reduces the heat current at off " state greatly but affects the heat current at on " state only slightly with the switching of about 1300 , where @xmath33 equals to 2.15x@xmath34 and 2.82x@xmath35 at @xmath28=0.01 and 0.14 , respectively .
it can be noticed that for this and the former lattice choice in the present model , @xmath36 at the off " and on " state is about @xmath34 and @xmath35 , respectively , with a switching of about @xmath44 , this is a significant improvement compared with previous one @xcite .
the high switching is mainly a result of the significant reduction of @xmath36 at the off " state , which is basically due to the lower @xmath45 at `` off '' state ( in the previous work it is designed at 0.04 ) .
moreover , we found that using lattice with lower heat resistance as terminal _
d _ is ideal for achieving a large @xmath37 . in fig .
2(b ) , we show the heat modulating effect with the same setup .
since the demonstrated features of the transistor rely very much on an efficient control of the interface temperature , it would be interesting also to study how the property of the control terminal can affect its functions . in fig .
3 , we show that when we use the fpu-@xmath41 lattice as the control segment , due to the large heat resistance in the segment , temperature @xmath45 does not well follow @xmath46 thus heat currents in segment @xmath1 and @xmath2 do not sensitively depend on @xmath46 in the low temperature regime . only when @xmath28@xmath47 0.08 , significant increment of currents and continuous modulation of @xmath36 are observed . in other words , @xmath36 can be even more precisely adjusted
. lattice with @xmath22(=@xmath48)=0.20 and @xmath41=0.05 is used as segment _
d_. other parameters remain the same as in fig . 1 ( b ) .
( b ) thermal modulator : all parameters remain the same as in 2 ( a ) except with @xmath4=0.25 , @xmath23=0.055 , and @xmath26=0.04 . ] and the fpu-@xmath41 lattices are used as segment @xmath49 and @xmath50 .
the system parameters are @xmath51=0.01 , @xmath18=5.00 , @xmath19=0.35 , @xmath4=0.60 ; @xmath20=0.20 , @xmath52=0.20 , @xmath53=0.05 , @xmath23=0.05 ; @xmath54=0.20 , @xmath55=2.00 , @xmath26=0.20 , and @xmath27=5.00 . ]
the results presented so far have shown that by choosing the lattice in different segment appropriately , one can improve the switching and modulating efficiency significantly .
this message is very important and useful for optimization and experimental realization of the thermal transistor . in the following
, we shall extend the study of the thermal transistor to 2d .
such extension is important in application , as it takes into account the interaction between longitudinal and transverse motions and therefore is closer to real material such as thin film _
etc_. however , it is not our intention here to look for optimum switching/ modulating condition for this model .
the displacement from equilibrium position and the momentum of a particle in 2d lattice are labeled as @xmath56 and @xmath57 , respectively .
the relative separation between two particles is @xmath58=@xmath59=@xmath60 .
the hamiltonian of our model consists of the hamiltonian of each segment : @xmath61,\end{aligned}\ ] ] and the hamiltonian of the interface : @xmath62.\end{aligned}\ ] ] similarly , all segments are coupled to each other via harmonic springs . in our model , segments @xmath63 and @xmath50 are fk lattices and segment @xmath49 is fpu-@xmath41 lattice .
thus , @xmath64(@xmath58)=@xmath65@xmath66 , @xmath67=@xmath68+@xmath69 , @xmath70=0 , and @xmath71=@xmath72 .
the harmonic spring constant @xmath12 and the on - site potential @xmath73 may take different values in different segments .
periodic and fixed boundaries are used in the y ( transverse ) and x ( longitudinal ) direction respectively , namely , @xmath74=@xmath75 and @xmath76=(0 , @xmath77 ) for @xmath77=@xmath78 .
thus the topologies of the segments , as shown in fig .
4(a ) , are cylinders .
nos - hoover heat baths are coupled to the ends of segment ( @xmath79=1 ) . in the stationary state ,
temperature gradients are set up in the @xmath80 direction and thus only heat currents through longitudinal links , @xmath81(=@xmath82 ) need to be taken into account , which are constant along each segment .
= 0.01 , @xmath19=0.20 , @xmath18=5.00 , @xmath4=0.50 ; @xmath20=0.17 , @xmath52=1.00 , @xmath41=3.00 , @xmath23=0.035 ; @xmath54=1.00 , @xmath24=6.00 , @xmath26=2.50 , and @xmath27=5.00 .
( c ) thermal modulator with @xmath51=0.015 . here , @xmath83=@xmath84=10 , @xmath85=9 , and @xmath86=4 .
] in fig .
4(b ) , we show the heat switching for the 2d thermal transistor model . it can be seen that when @xmath28 is at 0.01 and 0.07 , @xmath39@xmath30@xmath37 and @xmath87=0 . at these two points ,
@xmath36 are 8.43x@xmath88 and 3.38x@xmath35 respectively and thus the switching efficiency is about 4 .
this is a significant reduction when compared with the 1d case and this scenario is also quite similar to the 2d thermal rectifier @xcite .
we show in fig .
4(c ) that when @xmath51 is increased , the difference between @xmath39 and @xmath37 is reduced , which gives a better modulation of @xmath36 by the gate temperature . it is worth mentioning that in 2d case
, the crossover energy , particles above which are able to overcome the on - site potential barrier is much lower than that in 1d case .
thus the lattice parameters should be re - designed , otherwise the switching efficiency is reduced significantly as shown above .
in summary , we have provided several ways in improving the switching efficiency in 1d thermal transistor , namely , one can change different lattice in different segments , and/or by adjusting the coupling strength at the interface .
our study has been extended to two - dimensional lattices and a prototype of 2d thermal transistor - a tube - like model have been proposed .
our numerical simulations have demonstrated that such a prototype can work as a heat switch and a heat modulator - two basic functions of a thermal transistor . in our numerical simulations , we have used dimensionless units for the sake of computations .
however , all dimensionless units are related to the dimensional physical variables through dimensional scaling and consequently , the physical ( real ) temperature @xmath89 is related to the numerical temperature _ t _ through @xmath90 where @xmath91 is the boltzmann constant ( please refer to @xcite for details ) .
suppose we use single - walled carbon nanotube as our experimental prototype for the thermal transistor model , namely with the mass of particle @xmath92 , lattice constant @xmath93 , and natural oscillation frequency @xmath94 correspond to the real values of such a material @xcite , @xmath95=0.01 in our simulations corresponds to physical temperature @xmath89 of about @xmath96 k. similar analysis for heat flux gives @xmath97 ( = @xmath98 ) @xmath99@xmath88@xmath100 .
therefore , with @xmath100 @xmath99@xmath35 , @xmath97 is about @xmath101 w ( j / s ) which is smaller than that one in ref.@xcite .
the smaller heat flux obtained in our model is not surprising .
it is mainly due to the interacting with external on - site potential @xcite and the interface thermal resistance arises in coupling dissimilar nonlinear lattices @xcite , which is similar to the case of electronic diode where the electrical current are greatly reduced by the p - n junction . for a homogeneous fpu lattice with similar parameters , @xmath97 can be about 2 orders of magnitude higher . even though a real working prototype for solid state thermal transistor is still an open issue , we believe that the above model or its variants will be realized sooner or later
. moreover , a solid state thermal rectifier using nanotubes has been experimentally demonstrated @xcite ( see also ref@xcite for other variants ) .
more recently , a tunable and reversible thermal link using multi - walled carbon nanotubes ( mwcnts ) @xcite were successfully demonstrated whereby the thermal conductance can be controlled mechanically by displacing the outer shell with respect to the inner one .
this is a step forward to the realization of thermal transistor .
in fact , in the heat control models , be it a thermal rectifier or a thermal transistor , the key factors are : ( 1 ) the broken of the spatial symmetry of terminal , namely they are two different materials ; ( 2 ) nonlinearity in each segment .
therefore , the possible material to be used for the thermal transistor might be multiwalled nanotubes . as the recent experiment @xcite shows that the outer wall can be used to control the thermal conductance of the nanotube , thus it can be used as the controlling terminal g.
the work is supported partially by an academic research fund , r-144 - 000 - 203 - 112 from moe of singapore , and the temasek young investigator award of dsta singapore under project agreement pod0410553 199 f. bonetto , j. l. lebowitz , and l. rey - bellet : _ mathematical physics 2000 _ , edited by a. fokas , a. grigoryan , t. kibble , and b. zegarlinsky , ( imperial college press , london , 2000 ) , ( pp .
128 - 150 ) ; b. li , j. wang , l. wang , and g. zhang : chaos * 15 * ( 2005 ) 015121 .
m. terraneo , m. peyrard , and g. casati : phys .
88 * ( 2002 ) 094302 .
b. li , l. wang , and g. casati : phys .
* 93 * ( 2004 ) 184301 .
b. li , j. lan , and l. wang : phys .
* 95 * ( 2005 ) 104302 .
j. lan and b. li : phys .
b * 74 * ( 2006 ) 214305 .
d. segal and a. nitzan : phys .
* 94 * ( 2005 ) 034301 .
b. hu , l. yang , and y. zhang : phys .
* 97 * ( 2006 ) 124302 .
b. hu and l. yang : chaos * 15 * , 015119 ( 2005 ) .
n. yang , n. li , l. wang , and b. li : phys .
b * 76 * ( 2007 ) 020301 ( r ) .
b. li , l. wang , and g. casati : appl .
* 88 * ( 2006 ) 143501 . o .-
saira , m. meschke , f. giazotto , a. m. savin , m. mttnen , j. p. pekola : phys .
* 99 * ( 2007 ) 027203 .
l. wang and b. li : phys . rev
* 99 * ( 2007 ) 177208
. o. m. braun and y. s. kivshar : phys .
* 306 * ( 1998 ) 1 . v.
l. pokrovsky and a. l. talapov : _ theory of incommensurate crystals _ , soviet scientific reviews supplement series , physics vol . 1 ( harwood , new york , 1984 )
. s. nos : j. chem . phys .
* 81 * ( 1984 ) 511 ; w. g. hoover , phys .
a * 31 * ( 1985 ) 1695 .
k. r. patton and m. r. geller : phys .
b * 64 * ( 2001 ) 155320 .
b. hu , b. li , and h. zhao : phys .
e * 57 * ( 1998 ) 2992 .
k. aoki and d. kusnezov : phys .
* 86 * ( 2001 ) 4029 ; h. kaburaki and m. machida , phys .
a * 181 * ( 1993 ) 85 . for carbon atom , with the mass @xmath92=1.98x@xmath102 kg , lattice constant @xmath93@xmath103@xmath104 m , and the natural oscillation frequency @[email protected]@xmath105s@xmath106 ( if we take for example e=30mev @xcite ) , @[email protected]@xmath107@xmath95 and @xmath97 @xmath991.84x@xmath88@xmath100 .
g. zhang and b. li : j. chem . phys .
* 123 * ( 2005 ) 114714 .
for a ( 9,0 ) single - walled carbon nanotube whose ends are thermalized with nos - hoover heat baths at @xmath89=310k and 290k , @xmath97 is about 5.26x@xmath34 w. with roughly 4 carbon atoms per cross section of the nanotube , @xmath97 can be further approximated to about @xmath34 w. d. donadio and g. galli : phys .
* 99 * ( 2007 ) 255502 . c. w. chang , d. okawa , a. majumdar , and a. zettl : science * 314 * ( 2006 ) 1121 .
g. wu and b. li : phys .
b * 76 * ( 2007 ) 085424 . c. w. chang , d. okawa , h. garcia , t. d. yuzvinsky , a. majumdar , and a. zettl : appl . phys
90 * ( 2007 ) 193114 .
s. rols , z. benes , e. anglaret , j. l. sauvajol , p. papanek , j. e. fischer , g. coddens , h. schober , and a. j. dianoux : phys .
* 85 * ( 2000 ) 5222 . | thermal transistor is an efficient heat control device which can act as a heat switch as well as a heat modulator . in this paper
, we study systematically one - dimensional and two - dimensional thermal transistors .
in particular , we show how to improve significantly the efficiency of the one - dimensional thermal transistor .
the study is also extended to the design of two - dimensional thermal transistor by coupling different anharmonic lattices such as the frenkel - kontorova and the fermi - pasta - ulam lattices .
analogy between anharmonic lattices and single - walled carbon nanotube is drawn and possible experimental realization with multi - walled nanotube is suggested . |
the embedded contact homology ( ech ) of a closed , oriented @xmath3-manifold with a contact form was introduced by hutchings in @xcite and is a variant of the symplectic field theory @xcite of eliashberg , givental and hofer . it is defined in terms of a contact form but is an invariant of the underlying @xmath3-manifold .
this invariance has been established by taubes in @xcite via the identification with seiberg - witten floer ( co-)homology as defined by kronheimer and mrowka @xcite and in particular implies the weinstein conjecture in dimension three .
ech is also conjecturally isomorphic to ozsvth
szab heegaard floer homology defined in @xcite .
we would like to mention that kutluhan , lee and taubes , and independently colin , ghiggini and honda have recently announced two different proofs of the isomorphism between hutchings s embedded contact homology and heegaard floer homology .
a natural condition to impose on a compact , oriented contact @xmath4-manifold @xmath5 with boundary is to require that @xmath6 be _ convex _ , i.e. , there is a contact vector field @xmath7 transverse to @xmath8 . to a transverse contact vector field
@xmath7 we can associate the _ dividing set _
@xmath9 , namely the set of points @xmath10 such that @xmath11 . by the contact condition
, @xmath12 is a @xmath13-dimensional contact submanifold of @xmath14 ; the isotopy class of @xmath15 is independent of the choice of @xmath7 .
we will denote by @xmath16 the contact manifold @xmath5 with convex boundary and dividing set @xmath17 with respect to some transverse contact vector field @xmath7 .
note that the actual boundary condition we need is slightly different and is called a _ sutured boundary condition_. ( in the early 1980 s , gabai developed the theory of sutured manifolds @xcite , which became a powerful tool in studying @xmath3-manifolds with boundary . ) for the moment we write @xmath18 to indicate either the convex boundary condition or the sutured boundary condition .
it turns out that there is a way to generalize embedded contact homology to sutured @xmath3-manifolds .
this is possible by imposing a certain convexity condition on the contact form .
this construction is completely described in the paper of colin , ghiggini , honda and hutchings @xcite .
heegaard floer homology also admits a sutured version , namely the sutured floer homology ( sfh ) of juhsz @xcite , which is an invariant of sutured manifolds .
finally , kronheimer and mrowka in @xcite introduced the sutured version of seiberg - witten floer homology . extending the conjectured equivalence of heegaard floer homology and embedded contact homology , the following conjecture was formulated in @xcite : [ conj_echsfh ] @xmath19 , where @xmath20 denotes the relative spin@xmath21-structure determined by @xmath22 and @xmath23 . in this paper , we construct sutured contact solid torus with @xmath2 parallel longitudinal sutures , where @xmath24 , using the gluing method of colin , ghiggini , honda and hutchings @xcite and calculate the sutured embedded contact homology of it .
we apply the gluing method in such a way that the constructed sutured solid torus is equipped with a nondegenerate contact form satisfying the property that all closed embedded reeb orbits are noncontractible , define the same homology class and have the same symplectic action .
it turns out that for the constructed sutured manifolds the sutured version of embedded contact homology coincides with sutured floer homology . the corresponding calculation in sutured floer homology
has been done by juhsz in @xcite .
so far , this is the first series of nontrivial examples where these two theories provide the same answer .
[ suturedech ] let @xmath25 be a sutured manifold , where @xmath1 is a set of @xmath2 parallel longitudinal curves and @xmath24 .
then there is a contact form @xmath26 which makes @xmath27 a sutured contact manifold and @xmath28 where @xmath29 and @xmath30 is an adapted almost complex structure .
hence , @xmath31 there is a floer - type invariant of a closed , oriented contact odd - dimensional manifold , called contact homology .
contact homology was introduced by eliashberg and hofer and is a special case of the symplectic field theory . in @xcite , colin , ghiggini , honda and hutchings generalized contact homology to sutured manifolds . for the sutured contact manifold from theorem [ suturedech ] , we calculated the sutured versions of cylindrical contact homology and contact homology .
[ suturedcylconthom ] let @xmath32 be a sutured contact manifold from theorem [ suturedech ] .
then @xmath33 is defined , is independent of the contact form @xmath26 for the given contact structure @xmath22 and almost complex structure @xmath30 , @xmath34 and hence @xmath35 where @xmath36 is the homological grading and @xmath37 is the conley - zehnder grading .
[ suturedconthom ] let @xmath32 be a sutured contact manifold from theorem [ suturedech ] .
then @xmath38 is defined , is independent of the contact form @xmath26 for the given contact structure @xmath22 and almost complex structure @xmath30 , @xmath39 and hence @xmath40 where @xmath36 is the homological grading , @xmath37 is the conley - zehnder grading and @xmath41 denotes the coefficient of @xmath42 in the generating function @xmath43 .
this paper is organized as follows : in section [ background ] we review definitions of embedded contact homology , cylindrical contact homology and contact homology for sutured contact manifolds ; section [ construction ] describes the construction of sutured contact solid torus with @xmath2 longitudinal sutures , where @xmath24 ; finally in section [ calculations ] we calculate the relative versions of embedded contact homology , cylindrical contact homology and contact homology of the solid torus constructed in section [ construction ] .
the author is deeply grateful to ko honda for his guidance , help and support far beyond the call of duty .
he also thanks oliver fabert , jian he and michael hutchings for helpful suggestions and interest in his work .
in addition , the author is grateful to the referee of an earlier version of this paper for many valuable comments and suggestions .
the goal of this section is to review definitions of embedded contact homology , cylindrical contact homology and contact homology for contact sutured manifolds .
this section is essentially a summary of @xcite . in this section
we repeat some definitions from @xcite .
a _ liouville manifold _
( often also called a liouville domain ) is a pair @xmath44 consisting of a compact , oriented @xmath2-dimensional manifold @xmath45 with boundary and a @xmath46-form @xmath47 on @xmath45 , where @xmath48 is a positive symplectic form on @xmath45 and the _ liouville vector field _ @xmath49 given by @xmath50 is positively transverse to @xmath51 .
it follows that the @xmath46-form @xmath52 ( this notation means @xmath47 pulled back to @xmath51 ) is a positive contact form with kernel @xmath53 .
a compact oriented @xmath54-dimensional manifold @xmath55 with boundary and corners is a _ sutured manifold _ if it comes with an oriented , not necessarily connected submanifold @xmath56 of dimension @xmath57 ( called the _ suture _ ) , together with a neighborhood @xmath58\times[-1 , 1]\times \gamma$ ] of @xmath59 in @xmath55 , with coordinates @xmath60\times [ -1,1]$ ] , such that the following holds : * @xmath61\times \gamma)\cup([-1,0]\times \{-1\ } \times \gamma)\cup ( [ -1,0]\times \{1\}\times \gamma)$ ] ; * @xmath62 , where the orientation of @xmath8 agrees with that of @xmath63 and is opposite that of @xmath64 , and the orientation of @xmath1 agrees with the boundary orientation of @xmath63 ; * the corners of @xmath55 are precisely @xmath65 . the submanifold @xmath66 is called the _ horizontal boundary _ and @xmath67\times \gamma$ ] the _ vertical boundary _ of @xmath55 .
let @xmath68 be a sutured manifold .
if @xmath22 is a contact structure on @xmath55 ( this means that @xmath55 is now @xmath69-dimensional ) , we say that @xmath70 is a _ sutured contact manifold _ if @xmath22 is the kernel of a positive contact 1-form @xmath26 such that : * @xmath71 and @xmath72 are liouville manifolds ; * @xmath73 inside @xmath74 , where @xmath75 and @xmath47 is independent of @xmath76 and does not have a @xmath77-term ; * @xmath78 , where @xmath79 is a liouville vector field for @xmath80 .
such a contact form is said to be _ adapted _ to @xmath68 .
let @xmath81 be a sutured contact manifold with an adapted contact form @xmath26 .
the form @xmath26 is then given by @xmath82 on the neighborhoods @xmath83\times r_{+}(\gamma)$ ] and @xmath84\times r_{-}(\gamma)$ ] of @xmath85 and @xmath86 , where @xmath87\cup [ 1-\varepsilon,1]$ ] extends the @xmath76-coordinate on @xmath88 .
on @xmath88 , @xmath89 , @xmath90 , and @xmath91 is a liouville vector field @xmath49 for @xmath47 .
following the procedure explained in @xcite we can complete " @xmath92 to a noncompact contact manifold @xmath93 .
we first extend @xmath26 to @xmath94 and @xmath95\times r_{-}(\gamma)$ ] by taking @xmath96 as appropriate .
the boundary of this new manifold is @xmath97 .
notice that since @xmath98 , the form @xmath99\times \{t\}\times \gamma}$ ] is the symplectization of @xmath100 in the positive @xmath101-direction .
we glue @xmath102 with the form @xmath103 , where @xmath104 is the pullback of @xmath47 to @xmath105 .
let @xmath106 be the noncompact extension of @xmath55 described above and @xmath107 be the extension of @xmath26 to @xmath106 .
for convenience , we extend the coordinates @xmath108 so far defined only on the ends of @xmath106 to functions on @xmath106 so that @xmath109 $ ] and @xmath110 $ ] .
we then say that @xmath111 corresponds the top ( t ) , @xmath112 corresponds to the bottom ( b ) , and @xmath113 corresponds to the side ( s ) .
let @xmath114 be the extension / completion of @xmath115 , obtained by extending to ( s ) .
let @xmath116 be a sutured contact manifold with an adapted contact form @xmath26 and @xmath93 be its completion . the _ reeb vector field _
@xmath117 that is associated to a contact form @xmath107 is characterized by @xmath118 a _ reeb orbit _ is a closed orbit of the reeb flow , i.e. , a smooth map @xmath119 for some @xmath120 such that @xmath121 .
[ reeborbincompl ] every periodic orbit of @xmath117 lies in @xmath55 .
hence , the set of periodic reeb orbits of @xmath122 coincides with the set of periodic reeb orbits of @xmath123 .
consider reeb orbit @xmath124 passing through a point @xmath125 .
the linearization of the reeb flow on the contact planes along @xmath124 determines a linearized return map @xmath126 .
this linear map is symplectic and it does not depend on @xmath127 ( up to conjugation ) .
the reeb orbit @xmath124 is _ nondegenerate _ if @xmath128 .
note that nondegeneracy can always be achieved by a small perturbation , i.e. , for any contact structure @xmath22 on @xmath55 , there exists a contact form @xmath26 for @xmath22 such that all closed orbits of @xmath123 are nondegenerate . for simplicity , we assume that all reeb orbits of @xmath123 , including multiply covered ones , are nondegenerate . a reeb orbit @xmath124 is called _ elliptic _ or _ positive _ ( respectively _ negative _ ) _ hyperbolic _ if the eigenvalues of @xmath129 are on the unit circle or the positive ( resp .
negative ) real line respectively .
if @xmath101 is a trivialization of @xmath22 over @xmath124 , we can then define the conley - zehnder index .
in @xmath3-dimensional situation this is given explicitly as follows : if @xmath124 is elliptic , then there is an irrational number @xmath130 such that @xmath129 is conjugate in @xmath131 to a rotation by angle @xmath132 , and @xmath133 where @xmath132 is the total rotation angle with respect to @xmath101 of the linearized flow around the orbit .
if @xmath124 is positive ( respectively negative ) hyperbolic , then there is an even ( respectively odd ) integer @xmath134 such that the linearized flow around the orbit rotates the eigenspaces of @xmath129 by angle @xmath135 with respect to @xmath101 , and @xmath136 in this section we repeat some definitions from section 3.1 in @xcite .
let @xmath5 be a contact manifold with a contact form @xmath26 such that @xmath137 .
an almost complex structure @xmath30 on the symplectization @xmath138 is _ @xmath26-adapted _ if @xmath30 is @xmath139-invariant ; @xmath140 with @xmath141 for nonzero @xmath142 ; and @xmath143 , where @xmath144 denotes the @xmath145-coordinate and @xmath123 is a reeb vector field associated to @xmath26 .
let @xmath44 be a liouville manifold and @xmath53 be the contact structure given on @xmath51 by @xmath146 , where @xmath52 .
in addition , let @xmath147 be the completion of @xmath44 , i.e. , @xmath148 and @xmath149 , where @xmath101 is the @xmath150-coordinate .
an almost complex structure @xmath151 on @xmath152 is _
@xmath153- adapted _ if @xmath151 is @xmath104-adapted on @xmath154 ; and @xmath155 for all nonzero tangent vectors @xmath156 on @xmath45 .
let @xmath116 be a sutured contact manifold , @xmath26 be an adapted contact form and @xmath93 be its completion .
we say that an almost complex structure @xmath30 on @xmath157 is _ tailored to _
@xmath158 if the following hold : * @xmath30 is @xmath107-adapted ; * @xmath30 is @xmath159-invariant in a neighborhood of @xmath160 ; * the projection of @xmath30 to @xmath161 is a @xmath162-adapted almost complex structure @xmath151 on the completion @xmath163 of the liouville manifold @xmath164 .
moreover , the flow of @xmath159 identifies @xmath165 and @xmath166 .
first , let @xmath55 be a closed , oriented @xmath3-manifold , @xmath26 be a contact @xmath46-form on @xmath55 and let @xmath30 be an @xmath26-adapted almost complex structure on @xmath138 . for simplicity , we assume that all reeb orbits of @xmath123 , including multiply covered ones , are nondegenerate . an _ orbit set _ is a finite set of pairs @xmath167 , where the @xmath168 s are distinct embedded orbits of @xmath123 and the @xmath169 s are positive integers .
the orbit set @xmath170 is _ admissible _ if @xmath171 whenever @xmath172 is hyperbolic .
the homology class of @xmath170 is defined by @xmath173:=\sum\limits_{i } m_i[\alpha_i ] \in h_1(m ; \mathbb z).\end{aligned}\ ] ] if @xmath174 and @xmath175 are two orbit sets with @xmath176=[b]$ ] , let @xmath177 denote the set of relative homology classes of @xmath178-chains @xmath179 in @xmath55 with @xmath180 the _ ech chain complex _
@xmath181 is a free @xmath182-module with one generator for each admissible orbit set @xmath170 with @xmath176=h$ ] . if @xmath183 and @xmath184 are orbit sets with @xmath176 = [ b]$ ] , let @xmath185 denote the moduli space of @xmath30-holomorphic curves @xmath186 with positive ends at covers of @xmath172 with total multiplicity @xmath187 , negative ends at covers of @xmath188 with total multiplicity @xmath189 , and no other ends .
note that the projection of each @xmath190 to @xmath55 has a well - defined relative homology class @xmath191\in h_{2}(m , a , b)$ ] .
for @xmath192 we then define @xmath193 = z\}.\end{aligned}\ ] ] if @xmath174 is an orbit set , define the _ symplectic action _ @xmath194 [ le1_symplactfech ] for an adapted almost complex structure @xmath30 , if @xmath195 is non - empty , then : * @xmath196 . *
if @xmath197 , then @xmath198 and every element of @xmath199 maps to a union of trivial cylinders .
if @xmath200 , define the _ ech index _
@xmath201 here
@xmath202 denotes the relative intersection pairing , which is defined in @xcite .
any @xmath30-holomorphic curve @xmath203 can be uniquely written as @xmath204 , where @xmath205 and @xmath206 are unions of components of @xmath186 , each component of @xmath205 maps to an @xmath139-invariant cylinder , and no component of @xmath206 does .
[ pr1_echind ] suppose that @xmath30 is generic and @xmath207 .
then : * @xmath208 with equality if and only if @xmath209 . * if @xmath210 , then @xmath186 contains one embedded component @xmath206 with @xmath211 and @xmath205 does not intersect @xmath206 . to fix the signs in the differential ,
fix some ordering of all the embedded positive hyperbolic reeb orbits in @xmath55 .
two curves @xmath186 and @xmath212 in @xmath213 are equivalent if their embedded components @xmath214 and @xmath215 are the same up to translation , and if their other components cover each embedded trivial cylinder @xmath216 with the same total multiplicity .
the differential @xmath217 in @xmath218 counts @xmath219 curves in @xmath220 where @xmath170 and @xmath221 are admissible orbit sets .
such curves may contain multiple covers of the @xmath139-invariant cylinder @xmath216 when @xmath124 is an elliptic embedded reeb orbit .
the differential @xmath217 only keeps track of the total multiplicity of such coverings for each @xmath124 .
finiteness of the count results from the ech compactness theorem ( * ? ? ?
* lemma 9.8 ) . for the sign of the count
we refer to @xcite .
let @xmath116 be a sutured contact @xmath3-manifold with an adapted contact form @xmath26 , @xmath93 be its completion and @xmath30 be an almost complex structure on @xmath222 which is tailored to @xmath93 .
the sutured embedded contact homology group @xmath223 is defined to be the embedded contact homology of @xmath224 .
the following theorems have been proven by colin , ghiggini , honda and hutchings in @xcite : the ech compactness theorem ( * ? ? ?
* lemma 9.8 ) holds for @xmath30-holomorphic curves in the symplectization of the completion of a sutured contact @xmath3-manifold , provided that we choose the almost complex structure @xmath30 on @xmath225 to be tailored to @xmath93 .
let @xmath116 be a sutured contact @xmath3-manifold with an adapted contact form @xmath26 , @xmath93 be its completion and @xmath30 be an almost complex structure on @xmath222 which is tailored to @xmath93 .
then the embedded contact homology group @xmath226 is defined .
lemma [ le1_symplactfech ] and proposition [ pr1_echind ] hold for @xmath30-holomorphic curves in the symplectization of the completion of a sutured contact manifold , provided that we choose the almost complex structure @xmath30 on @xmath225 to be tailored to @xmath93 .
recall that embedded contact homology is an invariant of the underlying closed , oriented @xmath3-manifold .
hence , it is natural to expect the following : the embedded contact homology group @xmath227 does not depend on the choice of contact form @xmath26 , contact structure @xmath137 , and almost complex structure @xmath30 .
let @xmath116 be a sutured contact manifold with an adapted contact form @xmath26 , @xmath93 be its completion and @xmath30 be an almost complex structure on @xmath222 which is tailored to @xmath93 . for simplicity , we assume that all reeb orbits of @xmath123 , including multiply covered ones , are nondegenerate .
let @xmath124 be an embedded reeb orbit .
we are also interested in the multiple covers @xmath228 of @xmath124 , @xmath229 .
there are 2 ways the conley - zehnder index of @xmath230 can behave : * the parity of @xmath231 is the same for all @xmath232 . *
the parity for the even multiples @xmath233 , @xmath234 , disagrees with the parity for the odd multiples @xmath235 , @xmath236 . in the second case ,
the even multiples @xmath237 , @xmath234 , are called _
bad orbits_. an orbit that is not bad is called _ good_. the _ sutured contact homology algebra _
@xmath238 is defined to be the contact homology of @xmath224 in the following sense : the contact homology chain complex @xmath239 is the free supercommutative @xmath240-algebra with unit generated by good reeb orbits , where the grading and the boundary map @xmath217 are defined in the usual way ( as in @xcite ) with respect to the @xmath107-adapted almost complex structure @xmath30 .
the homology of @xmath241 is the sutured contact homology algebra @xmath242 .
we define the _ sutured cylindrical contact homology group _ @xmath243 to be the cylindrical contact homology of @xmath244 .
the cylindrical contact homology chain complex @xmath245 is the @xmath240-module freely generated by all good reeb orbits , where the grading and the boundary map @xmath217 are defined as in @xcite with respect to the @xmath107-adapted almost complex structure @xmath30 .
the homology of @xmath245 is the sutured cylindrical contact homology group @xmath246 .
for our calculations we will need the following fact which is a consequence of lemma 5.4 in @xcite : [ energyconthom ] let @xmath247 be a closed , oriented contact manifold with nondegenerate reeb orbits and @xmath248 be a @xmath30-holomorphic curve in @xmath249 , where @xmath124 and @xmath250 s are all good reeb orbits , @xmath30 is an @xmath26-adapted almost complex structure on @xmath138 and @xmath251 is a moduli space of @xmath30-holomorphic curves that we consider in contact homology . then @xmath252 with equality if and only if the image of @xmath253 is contained in a trajectory of @xmath123 , i.e. , @xmath186 maps to a trivial cylinder over @xmath254 , where @xmath254 is an embedded orbit of @xmath123 , and hence @xmath255 for some @xmath256 and @xmath257 with @xmath258 .
in addition , we recall the following fact proven by eliashberg and hofer : [ welldefcylconthom ] let @xmath247 be a closed , oriented contact manifold with nondegenerate reeb orbits .
let @xmath259 be the cylindrical contact homology complex , where @xmath36 is a homotopy class of reeb orbits and @xmath37 corresponds to the conley - zehnder grading .
if there are no contractible reeb orbits , then for every free homotopy class @xmath36 * @xmath260 ; * @xmath261 is independent of the contact form @xmath26 for @xmath22 , the almost complex structure @xmath30 and the choice of perturbation for the moduli spaces .
now we remind the following theorems which have been proven by colin , ghiggini , honda and hutchings in @xcite : the sft compactness theorem ( * ? ? ?
* theorem 10.1 ) holds for @xmath30-holomorphic curves in the symplectization of the completion of a sutured contact manifold , provided that we choose the almost complex structure @xmath30 on @xmath225 to be tailored to @xmath93 .
[ welldefconthom ] let @xmath116 be a sutured contact @xmath3-manifold with an adapted contact form @xmath26 , @xmath93 be its completion and @xmath30 be an almost complex structure on @xmath222 which is tailored to @xmath93 .
then the contact homology algebra @xmath262 is defined and independent of the choice of contact @xmath46-form @xmath26 with @xmath263 , adapted almost complex structure @xmath30 , and abstract perturbation .
[ actchomcchom ] fact [ energyconthom ] and fact [ welldefcylconthom ] hold for @xmath30-holomorphic curves in the symplectization of the completion of a sutured contact manifold , provided that we choose the almost complex structure @xmath30 on @xmath225 to be tailored to @xmath93 .
note that theorem [ welldefconthom ] and remark [ actchomcchom ] rely on the assumption that the machinery , needed to prove the analogous properties for contact homology and cylindrical contact homology in the closed case , works .
now we briefly describe the procedure of gluing sutured contact manifolds , together with compatible reeb vector fields which was first described by colin and honda in @xcite and generalized in @xcite .
in @xcite , gabai defined the notion of a sutured manifold decomposition for sutured 3-manifolds , which is the inverse construction of the sutured gluing .
let @xmath264 be a sutured contact @xmath3-manifold with an adapted contact form @xmath265 .
we denote by @xmath266 the projection along @xmath267 defined on @xmath268 .
if we think of @xmath269\times \gamma'$ ] as a subset of @xmath270 ( resp .
@xmath271 ) , then we denote the projection by @xmath272 ( resp .
@xmath273 ) . by definition ,
the horizontal components @xmath274 are liouville manifolds .
we denote by @xmath275 their liouville vector field .
the contact form @xmath265 is @xmath276 on the neighborhoods @xmath277 $ ] and @xmath278 $ ] of @xmath279 and @xmath280 .
in addition , we may assume without loss of generality that the reeb vector field @xmath281 is given by @xmath267 on @xmath268 .
take a @xmath178-dimensional submanifolds @xmath282 such that @xmath283 is the union of @xmath284 , @xmath285 and @xmath283 is positively transversal to the liouville vector field @xmath275 on @xmath286 . whenever we refer to @xmath287 and @xmath288
, we assume that closures are taken as appropriate .
moreover we make the assumption that @xmath289 .
let @xmath290 be a diffeomorphism which sends @xmath291 to @xmath292 and takes @xmath293 to @xmath294 and @xmath295 to @xmath296 .
we will refer to the triple @xmath297 as the _ gluing data_. for the purposes of gluing , we only need @xmath298 and @xmath299 to match up on @xmath300 , since we can linearly interpolate between primitives of positive area forms on a surface .
topologically , we construct the sutured manifold @xmath301 from @xmath302 and the gluing data @xmath297 as follows : let @xmath303 , where * @xmath304 for all @xmath305 ; * @xmath306 if @xmath307 and @xmath308 .
then @xmath309 and @xmath310 in dimension @xmath3 , for the purposes of studying holomorphic curves , we want to stretch in @xmath76-direction . in higher dimensions , one needs to stretch in both @xmath101- and @xmath76-directions .
the construction depends on the parameter @xmath311 , where @xmath311 is a stretching parameter in @xmath76-direction , and the resulting glued - up sutured contact manifold is written as @xmath312 .
let @xmath313 ( we will suppress @xmath311 to avoid cluttering the notation ) be the manifold obtained from the completion @xmath314 by removing the side ( s ) , i.e. , @xmath315).\end{aligned}\ ] ] then construct @xmath316 from @xmath317)),\end{aligned}\ ] ] by taking closures and identifying : * @xmath318 with @xmath319 ; * @xmath320 with @xmath321 ; * @xmath322 $ ] with @xmath323 $ ] ; all via the identification @xmath324 . next we take @xmath325 and truncate the top and bottom of @xmath316 to obtain the ( compact ) sutured manifold @xmath326 so that @xmath327 contains @xmath328)\cup ( ( \overline{r_{-}(\gamma')\setminus p_{-}})\times [ -n',-1]),\end{aligned}\ ] ] the reeb vector field is transverse to the horizontal boundary , and the vertical boundary @xmath329 is foliated by interval reeb orbits with fixed action @xmath330 .
attaching @xmath331\times e$ ] to @xmath327 for some specific @xmath332 gives us @xmath333 .
the horizontal boundary which is positively ( resp . negatively ) transverse to @xmath334 will be called @xmath335 ( resp .
@xmath336 ) . for more details
we refer to @xcite .
in this section we construct a sutured contact solid torus with @xmath2 longitudinal sutures , where @xmath24 .
now we construct @xmath337 .
the flow of the hamiltonian vector field associated to @xmath338 will play a role of gluing map when we will apply the gluing construction described in section [ section : glsutcontman ] to the sutured contact solid cylinder constructed in section [ section : gluing ] .
we fix @xmath339 and consider @xmath340 given by @xmath341 in polar coordinates about @xmath127 , where @xmath342 and @xmath343 .
note that @xmath344 is singular only at @xmath127 .
we obtain @xmath345 from @xmath344 by perturbing @xmath344 on a small disk @xmath346 about @xmath127 in such a way that @xmath338 has @xmath347 nondegenerate saddle points and interpolates with no critical points with @xmath344 on @xmath346 . in other words , @xmath348 on @xmath349 . for the level sets of @xmath344 and @xmath338 in the case
@xmath350 we refer to figure [ levelset3 ] .
the construction of @xmath338 was initially described by cotton - clay as a construction of a hamiltonian function whose time-@xmath46 flow is a symplectic smoothing of the singular representative of pseudo - anosov map in a neighborhood of a singular point with @xmath351 prongs in @xcite .
since some of the properties of @xmath338 described in @xcite will be important for further discussion , we will state them in the next remark .
[ re1_closedlevsets ] we can choose @xmath338 in such a way that it satisfies the following properties : * @xmath338 can be written as @xmath352 in some coordinates @xmath353 in a connected neighborhood containing its critical points ; * there are no components of level sets of @xmath338 which are circles ; * there is an embedded curve which is a component of one of the level curves of @xmath338 and connects all the saddle points of @xmath338 .
we call this embedded curve @xmath124 . for the detailed construction of @xmath338
we refer to lemmas 3.4 and 3.5 in @xcite .
( left ) and the level sets of @xmath338 ( right ) in the case @xmath350.,title="fig : " ] [ le1_prarfm ] let @xmath144 be a saddle point of @xmath338 .
then there are coordinates @xmath353 about @xmath144 such that @xmath354 for @xmath355 .
first observe that from remark [ re1_closedlevsets ] it follows that @xmath356 .
by morse lemma , there are coordinates @xmath357 about @xmath144 such that @xmath358 . given @xmath356 ,
we can write @xmath359 now let @xmath360 clearly @xmath127 and @xmath361 satisfy the statement of the lemma . in addition , the orientation of the pair @xmath353 coincides with the orientation of @xmath357 .
let @xmath362 be a neighborhood of the @xmath256-th saddle point @xmath363 of @xmath338 from lemma [ le1_prarfm ] . if @xmath364 is an exact symplectic manifold , i.e. , @xmath365 , then the flow @xmath366 of a hamiltonian vector field @xmath367 consists of exact symplectic maps , i.e. , @xmath368 since @xmath369 , @xmath370 since by definition @xmath371 , the integrand is equal to @xmath372\circ \varphi_{x_{h}}^{s}).\end{aligned}\ ] ] thus @xmath373 notice that our definition of @xmath367 is slightly different from the standard one ; usually @xmath367 is defined by @xmath374 .
note that the condition that @xmath375 is equivalent to the condition that @xmath376 .
[ re1_funcanalcon ] let @xmath377 .
in addition , let @xmath378 be a region such that @xmath375 on @xmath379 and @xmath380 \}$ ]
. then @xmath381 and @xmath382 on @xmath383 . in the next two lemmas
we construct a @xmath46-form @xmath47 on @xmath384 with @xmath385 and show that @xmath47 is `` adapted '' to @xmath338 , i.e. , @xmath386 near the saddle points of @xmath338 and on the region far enough from @xmath346 , where @xmath367 is a hamiltonian vector field with respect to @xmath387 and @xmath388 is the time-@xmath46 map of the flow of @xmath367 .
the condition that @xmath389 and remark [ re1_funcanalcon ] will play a crucial role when we will compare @xmath390 and @xmath47 .
[ le1_formanalcon ] there exists a 1-form @xmath47 on @xmath391 satisfying the following : * @xmath392 ; * the singular foliation given by @xmath393 has isolated singularities and no closed orbits ; * the elliptic points of the singular foliation of @xmath47 are the saddle points of @xmath338 ; @xmath394 on @xmath362 with respect to the coordinates from lemma [ le1_prarfm ] , where @xmath395 and @xmath396 is a small positive real number ; * @xmath397 on @xmath398 with respect to the polar coordinates whose origin is at the center of @xmath346 ; * the hyperbolic points of the singular foliation of @xmath47 are located on @xmath124 , outside of @xmath362 s and distributed in such a way that between each two closest elliptic points there is exactly one hyperbolic point . consider a singular foliation @xmath399 on @xmath391 which satisfies the following : * @xmath399 is morse - smale and has no closed orbits . * the singular set of @xmath399 consists of elliptic points and hyperbolic points .
the elliptic points are the saddle points of @xmath338 .
the hyperbolic points are located on @xmath124 and distributed in such a way that between each two closest elliptic points there is exactly one hyperbolic point .
in addition , the hyperbolic points are outside of @xmath362 s . * @xmath399 is oriented , and for one choice of orientation the flow is transverse to and exits from @xmath400 .
( left ) and the characteristic foliation of @xmath47 ( right ) in the case @xmath350,title="fig : " ] next , we modify @xmath399 near each of the singular points so that @xmath399 is given by @xmath401 on @xmath362 with respect to the coordinates from lemma [ le1_prarfm ] , and @xmath402 near a hyperbolic point . in addition , on @xmath349 , @xmath403 with respect to the polar coordinates whose origin is at the center of @xmath346 . finally , we get @xmath399 given by @xmath104 , which satisfies @xmath404 near the singular points and on @xmath349
. now let @xmath405 , where @xmath406 is a positive function with @xmath407 outside of @xmath408 , @xmath409 , @xmath410 and @xmath7 is an oriented vector field for @xmath399 ( nonzero away from the singular points ) .
since @xmath411 , @xmath412 guarantees that @xmath413 .
here @xmath396 is a small positive real number .
[ re1_formandfunc ] from the previous lemma we get @xmath47 defined on @xmath391 with the following properties : * @xmath414 on @xmath391 . * @xmath394 and @xmath354 on @xmath362 for @xmath415 . in other words ,
the saddle points of @xmath338 are exactly the elliptic singularities of @xmath47 .
* @xmath397 and @xmath416 on @xmath349 . for the comparison of the level sets of @xmath338 with the singular foliation of @xmath47 in the case @xmath350
we refer to figure [ charfol ] .
[ le2_formanalcon ] let @xmath47 be a @xmath46-form from lemma [ le1_formanalcon ] .
the hamiltonian vector field @xmath367 of @xmath338 with respect to the area form @xmath387 satisfies @xmath375 on @xmath417 .
first we work on @xmath362 , where @xmath395 . from remark
[ re1_formandfunc ] it follows that @xmath394 and @xmath354 on @xmath362 .
let @xmath367 be a hamiltonian vector field defined by @xmath418 .
we show that @xmath419 is a solution of the equation @xmath420 on @xmath362 .
we calculate @xmath421 and @xmath422 next , by remark [ re1_formandfunc ] , @xmath397 and @xmath423 on @xmath349 . as in the previous case , we show that @xmath424 is a solution of equation ( [ e1_formanconstr ] ) on @xmath398 .
we calculate @xmath425 and @xmath426 let @xmath427 be the time-@xmath144 flow of @xmath367 .
consider @xmath428 \}\quad \mbox{and}\\ & v_{k}:=\ { x\in u_{k } : \varphi_{x_{h}}^{s}(x)\in u_{k}\ \forall s\in[0,1 ] \}.\end{aligned}\ ] ] since the saddle points of @xmath338 are the fixed points of @xmath427 for @xmath429 $ ] , @xmath430 contains an open neighborhood about the @xmath256-th saddle point of @xmath338 .
in addition , note that @xmath379 contains the level sets of @xmath338 which do not intersect @xmath346 , and @xmath431 , where @xmath432 .
for ease of notation , we write @xmath388 instead of @xmath433 .
[ re1_analconstr ] lemma [ le2_formanalcon ] implies that @xmath434 on @xmath435 .
[ re1_phiperpoints ] from remark [ re1_closedlevsets ] and the fact that the flow of @xmath367 preserves the level sets of @xmath338 it follows that @xmath436 is the set of periodic points of @xmath388 on @xmath391 . in the next lemma
we construct a contact form @xmath26 on @xmath437 $ ] such that @xmath123 has vertical trajectories .
[ le1_almvertical ] let @xmath438 and @xmath439 be two @xmath46-forms on @xmath440 such that @xmath441 in a neighborhood of @xmath442 and @xmath443 then there exists a contact @xmath46-form @xmath26 with reeb vector field @xmath123 on @xmath444 \times d^2 $ ] with coordinates @xmath445 , where t is a coordinate on @xmath444 $ ] and @xmath127 is a coordinate on @xmath440 , with the following properties : * @xmath446 in a neighborhood of @xmath447 ; * @xmath448 in a neighborhood of @xmath449 ; * @xmath123 is collinear to @xmath450 on @xmath444\times d^2 $ ] ; * @xmath451 in a neighborhood of @xmath444\times \partial d^2 $ ] .
here @xmath396 is a small positive number .
since @xmath440 is simply connected and @xmath452 , there exists a function @xmath453 such that @xmath454
. let @xmath455\to [ 0,1]$ ] be a smooth map for which @xmath456 for @xmath457 , @xmath458 for @xmath459 and @xmath460 for @xmath461 $ ] , where @xmath462 is a small positive number .
in addition , we define @xmath463 .
consider @xmath444 \times d^2 $ ] equipped with a @xmath46-form @xmath464 we then compute @xmath465 and hence @xmath466 if @xmath396 is sufficiently small , then @xmath26 satisfies the contact condition , i.e. , @xmath467 .
now let us show that the reeb vector field @xmath123 is given by @xmath468 first we compute @xmath469 then we check the normalization condition , i.e. , @xmath470 : @xmath471 since @xmath472 in a neighborhood of @xmath473 , @xmath474 in a neighborhood of @xmath442 and hence @xmath451 in a neighborhood of @xmath444\times \partial d^2 $ ] . finally , we see that @xmath26 satisfies conditions @xmath475 .
fix @xmath476 such that there is an annular neighborhood @xmath477 of @xmath478 in @xmath384 with @xmath479 .
consider @xmath480 with two @xmath46-forms @xmath481 , where @xmath47 is a @xmath46-form from lemma [ le1_formanalcon ] , and @xmath482 . by remark [ re1_analconstr ] , @xmath483 in addition
, we have @xmath484 from equations ( [ e1_cylinf ] ) and ( [ e2_cylinf ] ) it follows that @xmath104 and @xmath485 satisfy the conditions of lemma [ le1_almvertical ] .
now take @xmath444\times d(r_{\ast})$ ] with the contact @xmath46-form @xmath26 from lemma [ le1_almvertical ] with @xmath104 and @xmath485 as in the previous paragraph .
note that @xmath486 for @xmath487 .
we can rewrite this equation as @xmath488 let us remind that @xmath489 on @xmath391 , where @xmath490 hence , @xmath491 satisfies equation ( [ e3_cylinf ] ) . from remark [ re1_funcanalcon ] it follows that @xmath492 on @xmath493 .
thus , @xmath474 on @xmath493 . since @xmath474 on @xmath494 , by the construction of @xmath26 , @xmath495 on @xmath496\times v_{k})\cup([-1,1]\times ( d(r_{\ast})\cap s))$ ] .
let @xmath497 and @xmath498 , where @xmath396 is a constant from lemma [ le1_almvertical ] which makes @xmath26 contact .
in this section we will construct the sutured contact solid torus with @xmath2 parallel longitudinal sutures , where @xmath24 .
first we construct surfaces with boundary @xmath499 with the following properties : * @xmath500 ; * @xmath501 and @xmath502 ; * @xmath388 maps @xmath503 to @xmath504 in such a way that @xmath505 and @xmath506 ; * @xmath507 .
recall that @xmath508 on @xmath509 .
note that @xmath367 is collinear to @xmath510 for @xmath511 , where @xmath512 . for simplicity ,
let us denote @xmath513 where @xmath514 .
fix @xmath334 such that @xmath515 and there is an annular neighborhood @xmath516 of @xmath517 in @xmath480 satisfying @xmath518 .
consider @xmath519 .
let @xmath520 be a segment on @xmath517 which starts at @xmath521 and ends at @xmath522 , i.e. , @xmath523 .
consider @xmath524 .
it is easy to see that every level set of @xmath338 which intersects @xmath520 intersects it only once .
hence , using that there are no closed level sets of @xmath338 and @xmath367 is @xmath525-symmetric on @xmath526 , we get @xmath527 let @xmath528 . by possibly making @xmath334 and
@xmath529 big enough , we can make @xmath530 s to be in @xmath516 .
consider the endpoints of @xmath530 s . since @xmath367 is @xmath525-symmetric outside of @xmath346 , it is easy to see that @xmath531 and @xmath532 , where @xmath533 , @xmath534 and @xmath535 .
in addition , observe that @xmath536 is the same for all endpoints of @xmath520 s .
let @xmath537 be a set of embedded curves on @xmath480 with the following properties : * @xmath538 starts at the terminal point of @xmath539 and ends at the initial point of @xmath520 , where @xmath256 is considered mod @xmath351 ; * @xmath540 and @xmath541 for @xmath542 ; * @xmath543 for @xmath542 ; * the region bounded by @xmath520 s and @xmath544 s has smooth boundary ; * each level set of @xmath338 which intersects @xmath544 intersects it only once . for simplicity , we take @xmath525-symmetric @xmath544 s , i.e.
, @xmath545 can be obtained from @xmath544 , where @xmath256 is considered mod @xmath351 , by doing @xmath525-positive rotation about the center of @xmath480 . , @xmath504 and @xmath546 in the case @xmath350,title="fig : " ] note that properties @xmath547 and @xmath548 and the form of @xmath367 on @xmath526 imply that @xmath549 where @xmath512 . again , using that the level sets of @xmath338 which intersects @xmath544 intersects it only once , there are no closed level sets of @xmath338 and @xmath367 is @xmath525-symmetric , we obtain @xmath550 let @xmath551 . from formula ( [ e1_infgl ] ) and the construction of @xmath520 s it follows that @xmath552 then we connect the terminal point of @xmath520 with the initial point of @xmath553 by the line segment @xmath554 and the terminal point of @xmath553 with the initial point of @xmath555 by the line segment @xmath556 . from the construction above
it follows that @xmath554 intersects @xmath557 only at the terminal point of @xmath520 , and @xmath556 intersects @xmath557 only at the initial point of @xmath555
. then we round the corners between @xmath554 and @xmath520 , @xmath554 and @xmath553 , @xmath556 and @xmath553 , @xmath556 and @xmath555 .
finally , we get a surface whose boundary consists of @xmath520 s , @xmath553 s , @xmath556 s and @xmath554 s , which we call @xmath546 .
see figure [ constofd ] .
[ re1_boundparts ] by the construction , @xmath558 s , @xmath559 s and @xmath560 lie in @xmath561 .
now we take @xmath444\times d$ ] with a contact form @xmath562\times d}$ ] and contact structure @xmath563 .
let @xmath564 in @xmath444\times d$ ] and @xmath565\times [ -1,1]\times \gamma$ ] be a neighborhood of @xmath1 with coordinates @xmath566\times [ -1,1]$ ] , where @xmath76 is a usual @xmath76-coordinate on @xmath444\times d$ ] . by remark [ re1_boundparts ]
, we can make @xmath74 such that @xmath567\times ( d\cap s).\end{aligned}\ ] ] observe that @xmath568\times d)=r_{+}\cup r_{-}$ ] , where @xmath569 and @xmath570 with respect to the coordinates on @xmath444\times d$ ] .
in addition , @xmath571\times d)=[-1,1]\times \partial d$ ] with respect the coordinates on @xmath444\times d$ ] .
let @xmath572 .
[ le1_suturbefglug ] @xmath573\times d , \gamma , u(\gamma ) , \xi)$ ] is a sutured contact manifold and @xmath26 is an adapted contact form . first note that @xmath574 and @xmath575 .
let us check that @xmath576 and @xmath577 are liouville manifolds . from the construction of @xmath80 it follows that @xmath578 .
since @xmath579 on @xmath580 , by equation ( [ e3_infgl ] ) , we have @xmath581 on @xmath74 . recall that @xmath582 on @xmath583 .
hence , @xmath584 .
the calculation @xmath585 implies that the liouville vector fields @xmath586 are equal to @xmath587 . from the construction of @xmath546
it follows that @xmath79 is positively transverse to @xmath588
. therefore , @xmath589 and @xmath590 are liouville manifolds .
as we already mentioned , @xmath581 on @xmath74 .
finally , if we take @xmath101 such that @xmath591 , then @xmath573\times d , \gamma , u(\gamma ) , \xi)$ ] becomes a sutured contact manifold with an adapted contact form @xmath26 .
now observe that from the construction of @xmath26 it follows that @xmath592\times d}=dt+\beta_{-}$ ] and @xmath593\times d}=dt+\beta_{+}$ ] .
then we define @xmath594 .
let @xmath503 be a region bounded by @xmath520 s and @xmath544 s in @xmath595 and let @xmath504 be a region bounded by @xmath530 s and @xmath553 s in @xmath596 .
note that from remark [ re1_boundparts ] it follows that @xmath558 s and @xmath559 are in @xmath379 .
hence , by lemma [ le1_suturbefglug ] , @xmath597 along @xmath283 .
the construction of @xmath558 s and @xmath559 implies that @xmath79 is positively transverse to @xmath283 . from the construction we made
it is easy to see that @xmath598 , @xmath599 , @xmath600 and @xmath601 . if @xmath602\times d \to d$ ]
is a projection to @xmath546 along @xmath267 , then from equation ( [ e2_infgl ] ) it follows that @xmath603 .
observe that @xmath604 and @xmath605 , @xmath606 for @xmath542 .
hence , by definition of @xmath607 , @xmath388 sends @xmath503 to @xmath504 in such a way that @xmath293 maps to @xmath294 and @xmath295 maps to @xmath296 .
next , we follow the gluing procedure overviewed in section [ section : glsutcontman ] and completely described in @xcite . we get a sutured contact manifold @xmath608 . for simplicity , we omit index @xmath311 . observe that the region enclosed by @xmath473 and @xmath530 in @xmath609 contains @xmath520 and the region enclosed by @xmath473 and @xmath544 in @xmath610 contains @xmath553 .
then from the gluing construction and the form of @xmath388 near the boundary of @xmath503 it follows that @xmath611 has @xmath2 parallel longitudinal components .
consider @xmath608 obtained in section [ section : gluing ] . recall that @xmath611 consists of @xmath2 parallel longitudinal curves
let @xmath612 denote the contact structure defined by @xmath613 and @xmath614 denote the reeb vector field defined by @xmath613 .
the main goal of this section is to understand the set of embedded , closed orbits of @xmath614 .
let @xmath379 be a non - empty set with two non - empty subsets @xmath615 and @xmath616 such that @xmath617 , and let @xmath618 .
a point @xmath619 is called a _ periodic point _ of @xmath253 of period @xmath351 if @xmath620 is well - defined , i.e. , @xmath621 for @xmath622 , and @xmath623 .
[ le1_anconstnumboforb ] @xmath614 has @xmath347 embedded , closed orbits .
first consider @xmath624 . recall that from the construction of @xmath504 and @xmath503 it follows that @xmath625 .
hence , by remark [ re1_phiperpoints ] , @xmath626 is the set of periodic points of @xmath624 . from the construction of @xmath26 on @xmath444\times d$ ] and the gluing construction it follows that there is a one - to - one correspondence between the set of embedded reeb orbits and the set of periodic points of @xmath624 .
thus , there are @xmath347 embedded closed orbits of @xmath614 .
let @xmath627 be the embedded , closed orbit , which corresponds to the periodic point @xmath363 , i.e. , @xmath627 is obtained from @xmath444\times \{p_{k}\}\subset [ -1,1]\times d$ ] .
[ le1_anconstsymplact ] @xmath628 is a nondegenerate orbit for @xmath629 and @xmath630 .
moreover @xmath631 is a set of positive hyperbolic orbits and @xmath632 for @xmath633 .
let @xmath634\times d)\cup ( r_{+}(\gamma)\times[1;\infty))\cup ( r_{+}(\gamma)\times ( -\infty;-1]))\end{aligned}\ ] ] and @xmath635 in addition , let @xmath636 denote the contact form on @xmath637 and let @xmath638 denote the contact structure defined by @xmath636 . consider @xmath444\times d\subset \tilde{m}$ ] . from the construction of @xmath26 it follows that @xmath639\times v_{k}}=dt+\beta_{-}$ ] .
since the contact structure on @xmath640 is given by @xmath641 and the contact structure on @xmath642\times p_{-}$ ] is given by @xmath643 , @xmath644\times v_{k}}=dt+\beta_{-}$ ] on @xmath645\times v_{k}\subset \tilde{m}$ ] .
therefore , we get @xmath646\times \{p_{k}\ } } \alpha_{\tilde{m}}=2n.\end{aligned}\ ] ] from the gluing construction and equation ( [ e10_anconstlength ] ) it follows that @xmath647 .
note that @xmath648 does not depend on @xmath256 .
hence , @xmath649 for @xmath633 .
now observe that @xmath650 and hence @xmath651 where @xmath652 .
let the symplectic trivialization of @xmath638 along @xmath645\times \{p_{k}\}$ ] be given by the framing @xmath653 .
note that the symplectic trivialization of @xmath638 gives rise to the symplectic trivialization of @xmath612 along @xmath627 .
it is easy to see that the linearized return map @xmath654 is given by @xmath655 since the eigenvalues of @xmath654 are positive real numbers different from @xmath46 , @xmath627 is a positive hyperbolic orbit .
hence , @xmath631 is a set of positive hyperbolic orbits of @xmath614 .
in addition , @xmath656 .
therefore , the eigenvalues of @xmath657 are different from @xmath46 .
hence , @xmath628 is a nondegenerate orbit for @xmath630 .
in this section we will calculate the sutured embedded contact homology , the sutured cylindrical contact homology and the sutured contact homology of the sutured contact solid torus constructed in section [ section : gluing ] .
consider the symplectization @xmath658 of @xmath659 , where @xmath144 is the coordinate on @xmath139 and @xmath660 is the completion of @xmath661 .
let @xmath30 be an almost complex structure on @xmath662 tailored to @xmath663 .
consider the set of embedded , closed orbits of @xmath614 . by lemma [ le1_anconstnumboforb ]
, @xmath614 has @xmath347 embedded , closed orbits @xmath664 , which are positive hyperbolic by lemma [ le1_anconstsymplact ] . in addition , lemma [ le1_anconstsymplact ] implies that all reeb orbits are nondegenerate . from the gluing construction ,
i.e. , since @xmath665 is a set of fixed points of @xmath388 , it follows that @xmath666 $ ] is a generator of @xmath667 for @xmath668 and @xmath666=[\gamma_{j}]$ ] for @xmath669 . from now on we identify @xmath670 with @xmath182 in such a way that @xmath666\in h_{1}(s^1\times d^2 ; \mathbb z)$ ] is identified with @xmath671 for @xmath672 .
recall that multiplicities of hyperbolic orbits in an admissible orbit set must be equal to @xmath46 .
hence , from lemma [ le1_anconstsymplact ] it follows that the admissible orbit sets are of the form @xmath673 , where @xmath674 .
note that @xmath675 is an admissible orbit set .
for ease of notation , we write @xmath676 instead of @xmath677 and @xmath46 instead of @xmath675 , where @xmath678
. fix @xmath680 .
let @xmath681 be a set of admissible orbit sets with homology class @xmath36 .
it is easy to see that @xmath682 from lemma [ le1_anconstsymplact ] it follows that for every @xmath683 , @xmath684 .
let @xmath685 be different admissible orbit sets . then , as we already mentioned , @xmath686 from equation ( [ e11_anconstdifforbss ] ) and the second part of lemma [ le1_symplactfech ] it follows that @xmath687 is empty .
in addition , by the second part of lemma [ le1_symplactfech ] , every element in @xmath688 maps to a union of trivial cylinders .
hence , by proposition [ pr1_echind ] and definition of @xmath217 , @xmath679 .
note that trivial cylinders are regular and hence we can omit the genericity assumption in proposition [ pr1_echind ] .
thus , for every admissible orbit set @xmath170 , @xmath679 . by counting the number of element in @xmath681
, we get @xmath689 by equation ( [ e9_acech ] ) and lemma [ le1_anconstechdiff ] , we get @xmath690 here @xmath691 is the exterior algebra over @xmath182 generated by @xmath664 .
thus , we obtain @xmath692 this completes the proof of theorem [ suturedech ] .
note that for the constructed sutured contact solid torus , the sutured floer homology coincides with the sutured embedded contact homology .
in fact , they agree in each spin@xmath21-structure .
this follows from proposition 9.2 in @xcite , where the sutured floer homology of every sutured manifold @xmath25 has been computed by juhsz .
[ cch_indep ] note that there are no contractible reeb orbits .
hence , from fact [ welldefcylconthom ] , remark [ actchomcchom ] and the fact that @xmath693 it follows that for all @xmath694 @xmath695 is defined , i.e. , @xmath260 , and is independent of contact form @xmath613 for the given contact structure @xmath612 and the almost complex structure @xmath30 . from lemma [ le1_anconstsymplact ]
it follows that @xmath631 is a set of positive hyperbolic orbits .
hence , the definition of the conley - zehnder index implies that @xmath698 is even for @xmath699 and @xmath630 .
then , according to the definition of a good orbit , it follows that @xmath700 is a good orbit for @xmath701 and @xmath630 .
hence , we get @xmath702 here @xmath703 is a @xmath240-module freely generated by @xmath704 .
now recall that lemma [ le1_anconstsymplact ] says that @xmath705 , where @xmath706
. therefore , @xmath707 for @xmath706 and @xmath630 .
remark [ actchomcchom ] , equation ( [ e21_anconsymact ] ) and definition of the cylindrical contact homology differential imply that @xmath708 for all @xmath694 .
thus , using equation ( [ e15_accch ] ) , we obtain @xmath709 finally , equation ( [ e22_anconstcch ] ) implies that @xmath710 this completes the proof of theorem [ suturedcylconthom ] .
recall that from lemma [ le1_anconstnumboforb ] it follows that all closed orbits of @xmath614 are nondegenerate . from the discussion in the previous section it follows that @xmath700 is a good orbit for @xmath699 and @xmath630 .
hence , the supercommutative algebra @xmath711 is generated by @xmath700 for @xmath701 and @xmath630 . note that @xmath711 splits as @xmath712 where @xmath713 is generated , as a vector space over @xmath240 , by monomials of total homology class @xmath36 .
hence , @xmath714 where @xmath715 denotes the coefficient of @xmath42 in the generating function @xmath43 . in ( * ? ? ?
* corollary 4.2 ) , fabert proved that the differential in contact homology and rational symplectic field theory is strictly decreasing with respect to the symplectic action filtration .
in other words , branched covers of trivial cylinders do not contribute to contact homology and rational symplectic field theory differentials . from lemma [ le1_anconstsymplact ]
it follows that all generators of @xmath713 have the same symplectic action and hence @xmath716 for all @xmath680 .
thus , @xmath717 and hence @xmath718 this completes the proof of theorem [ suturedconthom ] . | we calculate the relative versions of embedded contact homology , contact homology and cylindrical contact homology of the sutured solid torus @xmath0 , where @xmath1 consists of @xmath2 parallel longitudinal sutures . |
we have recently shown that in a rotating , turbulence - free and isothermal gaseous nebula with a density function that maximizes locally , the combined effect of gas drag and pressure gradients at hydrostatic equilibrium , causes solids to migrate toward the locations of local density enhancements ( * ? ? ?
* hereafter hb03 ) .
the motions of solids in hb03 were restricted to the midplane of the nebula and the variation of the gas density along the direction perpendicular to the midplane was neglected .
the results of our numerical simulations of the dynamics of micron - sized dust grains to 100 meter - sized objects indicated that the radial motions of such solids could be quite rapid when their sizes range from 1 cm to 1 m. in this paper , we extend our previous analysis to three - dimensional motions and study the radial and vertical migrations of solids subject to gas drag and the gravitational attraction of a solar type star in a non - uniform gaseous nebula .
study of the motions of solids in gaseous mediums has long been of particular interest to astronomers . because of the great diversity in the applicability of the results of such studies , in particular , to the process of formation of planetesimals ,
the literature in this area has become so rich that it makes it virtually impossible to cite all the related articles here . for a short review of the history of such studies
we refer the reader to hb03 and the references therein . of direct relevance to this study , however , are two classic papers by @xcite and @xcite , an article by @xcite on the study of the orbital evolution of solids in a turbulent protostellar disk with application to the formation of icy planetesimals in the outer region of the nebula and a recent paper by @xcite on the radial migration of dust grains in an accreting solar nebula .
our model nebula is turbulence - free and isothermal .
we assume that it consists of pure molecular hydrogen and follows the equation of state of an ideal gas .
an enhancement in the pressure of the gas in such a nebula appears where the density of the gas is locally enhanced . to focus attention on the rate of migration of solids and its dependence on their sizes and physical properties , the density of the nebula is assumed to have an azimuthally symmetric maximum on the midplane . along the vertical axis ,
the gas density changes in a way that the vertical component of the pressure gradients balances the vertical component of the gravitational force of the central star .
the direction of radial motions of solids in our model nebula depends on their initial positions relative to the location of the maximum gas density and their vertical migration is always toward the midplane .
@xcite .
we start by explaining our physical model in @xmath0 2 and present the equations of motions of solids in @xmath0 3 . in @xmath0 4 , we present the results of our numerical simulations and in @xmath0 5 , we conclude our study by reviewing the results and discussing their applications .
we consider a rotating , turbulence - free , and isothermal gaseous nebula of pure molecular hydrogen with a sun - like star at its center . at hydrostatic equilibrium and in a cylindrical coordinate system with the star at its origin and the midplane of the nebula as its polar plane , the gravitational attraction of the star at a point with a position vector @xmath1 is related to the pressure gradients of the gas as @xmath2 in @xmath3-direction and @xmath4 along the @xmath5-axis . in equations ( 1 ) and ( 2 ) ,
@xmath6 and @xmath7 are the density and pressure of the gas , respectively , @xmath8 is the angular velocity of the gas on a plane parallel to the midplane at a height @xmath5 , and @xmath9 is its corresponding keplerian angular velocity given by @xmath10 where @xmath11 is the gravitational constant and @xmath12 is the mass of the central star . the balance of the gravitational attraction of the central star by the pressure gradients in the vertical direction , as given by equation ( 2 ) , determines the distribution of the gas along the @xmath5-axis . in the model nebula
considered here , the gas is pure molecular hydrogen and it , therefore , obeys the equation state of an ideal gas . as a result
, the pressure and density of the gas are related as @xmath13 where @xmath14 is the boltzmann constant , @xmath15 is the temperature of the gas and @xmath16 is the molecular mass of hydrogen . substituting for @xmath7 from equation ( 4 ) in equation ( 2 )
, the density of our model nebula will be given by @xmath17\biggr\}.\ ] ] in equation ( 5 ) , @xmath18 is the density of the gas on the midplane . as mentioned in @xmath0 1 , in order to focus attention on the dynamics of solids , we consider a nebula with an azimuthally symmetric density enhancement on its midplane .
following hb03 , we assume @xmath19\,,\ ] ] where @xmath20 and @xmath21 are constant quantities .
figure 1 shows an edge - on view of our model nebula with the line of sight perpendicular to one of the midplane diameters .
the central star in this figure is solar - type and the gas temperature is 300 k. the values of @xmath22 and @xmath23 are equal to @xmath24 and 1 au , respectively .
the equation of motion of an object with mass of unity in our model nebula is given by @xmath25 where @xmath26 represents the drag force of the gas for the unit mass of a solid .
following @xcite and as presented in hb03 , we write @xmath26 as @xmath27\ , { { \bf v}_{\rm { rel}}}\,.\ ] ] in equation ( 8) , @xmath28 is the radius of the object , @xmath29 denotes its density , @xmath30 is the magnitude of its velocity relative to the gas and @xmath31 is the mean free path of the gas molecules . for a nebula of pure molecular hydrogen , @xmath32 g @xmath33 ( hb03 ) . the quantity @xmath34 in equation ( 8) is the mean thermal velocity of the gas molecules and @xmath35 is the drag coefficient whose numerical value is given by @xcite @xmath36 in equation ( 9 ) , @xmath37 is the gas reynolds number and is equal to @xmath38 where @xmath39 is the collisional cross section of hydrogen molecules .
the velocity of a solid relative to the gas is equal to @xmath40 where @xmath41 and @xmath42 represent the velocities of the solid and gas molecules , respectively . in the cylindrical coordinate system considered here , @xmath43 has the following components .
@xmath44 where we have assumed that the gas molecules do not move along the @xmath5-axis .
it is important to mention that it is the @xmath3 and @xmath5 components of @xmath43 that play important roles in the radial and vertical migrations of solids .
they also dominate the magnitude of @xmath43 when the object is small .
the transverse component of the relative velocity , @xmath45 , will have a dominating effect in the magnitude of @xmath43 for larger objects and it plays an important role in changing the angular momenta of such solids which can have significant contributions when accretion is considered . from equations ( 11 ) to ( 13 ) , the equation of motion of a solid with a unit mass can be written as @xmath46 in equations ( 14 ) to ( 19 ) , the hat sign indicates a dimensionless quantity .
we have followed our convention presented in hb03 and have chosen two quantities @xmath47 and @xmath48 to represent the units of length and time , respectively , in such a way that @xmath49 .
the quantities @xmath50 and @xmath51 are the dimensionless radial , vertical and angular momenta of the object , @xmath52 and @xmath53 .
to start integrating equations ( 14 ) to ( 19 ) , numerically , we take the mass of the central star to be equal to one solar mass and we choose , @xmath54 and @xmath55 g @xmath33 . as in hb03 , the azimuthal symmetry of the gas density on the midplane of our model nebula allows us to set @xmath56 au which implies @xmath48=0.16 years .
an important quantity that has to be determined at the beginning of our integrations is the drag coefficient , @xmath35 . as mentioned in the last section ,
this quantity is a function of the reynolds number of the gas as given by equation ( 9 ) . at the beginning of the integration , the initial functional form of @xmath35
is determined by calculating the initial value of @xmath37 . during the course of integration , the value of @xmath37
is constantly monitored .
once the range of this quantity changes , the integration is continued by adjusting the drag coefficient to its new functional form .
we integrated equations ( 14 ) to ( 19 ) for different values of solids radii and densities and also for different values of the gas temperature . the objects were initially placed on circular orbits at different radial positions .
the initial height @xmath5 of an object was taken to be one - tenth of its initial radial distance and @xmath57 . the initial velocity along the @xmath5-axis
was set equal to zero .
figures 2 and 3 show the radial migration of objects with radii ranging from 1 micron to 1 meter .
the densities of the objects are equal to 2 g @xmath33 and the temperature of the gas is 300 k. each object is once placed at ( 2,0.2 ) au with a radial distance larger than the radial location of the density enhancement on the midplane ( i.e. , @xmath58 au ) , and is once placed at a closer radial distance at ( 0.25,0.025 ) au . as expected , all objects migrate radially in / out toward @xmath58 au ( hb03 ) .
smaller objects show the tendency of staying with the gas and undergo very slow radial migrations .
larger objects , on the other hand , show more rapid motions . as in the case of the two - dimensional model ( hb03 )
, a meter - sized object undergoes the most rapid radial migration .
figures 2 and 3 also show that the rate of radial migration is greater for inward motions .
a result that was also observed in two - dimensional simulations of hb03 .
this can be attributed to the fact that the rate of change of angular momentum for an object beyond the radial location of the density enhancement is larger than that of an object in a closer distance to the central star . as a result ,
farther objects lose angular momentum faster , hence , their more rapid radial migration ( see hb03 for more detailed analysis ) .
as mentioned before , as a result of the combined effect of gas drag and pressure gradients , it is expected that objects at radial distances larger than @xmath59 to undergo inward radial migrations
. a closer investigation of the results shown in figures 2 and 3 , however , indicates that at the beginning of their radial motions , objects at @xmath60 au have , briefly , undergone outward migrations .
figure 4 shows this for a millimeter- and a centimeter - sized particle .
such a behavior can be understood by studying the structure of our model nebula . as given by equation ( 5 ) , the maxima of the gas density on planes parallel to the midplane at different heights @xmath5 will appear at different radial distances .
figure 5 shows this for @xmath5 ranging from zero to 0.2 au . in the top graph of this figure
we have plotted the density of the gas as a function of @xmath3 for constant values of @xmath5 . as shown here , the radius of the circular enhancement of the nebula increases from @xmath58 au on the midplane to larger values at higher vertical distances .
the middle and the bottom graphs of figure 5 show the gas density enhancement at @xmath61 0.15 and 0.2 au , respectively . when an object is initially at a radial distance beyond 1 au , depending on its height from the midplane , it may be inside the circle of the local density enhancement at that height . in such cases
, the object starts its radial motion by an outward migration while descending toward the midplane .
once it passes the height where the radial distance of the local maximum density is smaller than its instantaneous radial position , the radial motion of the object reverses and it starts its inward migration .
we also numerically integrated equations ( 14 ) to ( 19 ) for different values of a solid s density .
figure 6 shows the radial migrations of 1 cm and 1 m solids for different values of their densities .
the temperature of the gas is 300 k. similar to the results of two - dimensional simulations ( hb03 ) , increasing the solid s density results in a more rapid radial migration for a centimeter - sized or smaller particle .
however , it has a reverse effect on the motions of meter - sized and larger objects .
such different responses to the change of the density of a solid are the consequences of the contribution of gas drag in changing the rates of its radial migration .
recall that , as mentioned in @xmath0 3 , the radial component of the velocity of an object relative to the gas is the dominant factor in its radial motion . for a centimeter - sized or smaller object
, the magnitude of this component , given by equations ( 11 ) and ( 14 ) , also dominates the magnitude of the relative velocity of the object . increasing the density of an object while keeping its radius constant results in decreasing the magnitude of the drag force @xmath62 given by equation ( 8) .
such a decrease in the magnitude of @xmath62 , in turn , causes the rate of change of the radial component of the relative velocity of the object ( equation [ 17 ] ) to increase , and as a result the object migrates more rapidly .
the situation is not the same for meter - sized and larger objects . in such cases ,
it is the transverse component of @xmath43 ( equation [ 13 ] ) that dominates its magnitude .
that is , as opposed to small particles , the motion of a meter - sized or larger object is dominated by the rate of the change of its angular momentum . because the magnitude of @xmath62 decreases by increasing the solid s density , from equation ( 19 ) , a larger value for @xmath29 results in a smaller rate for the decrease / increase of the angular momentum of the object . in other words ,
an increase of the solid s density will increase its reluctance in loosing / gaining angular momentum and hence a decrease in its rate of inward / outward radial migrations .
numerical integrations were also carried out for different values of the gas temperature . figure
7 shows radial migrations of a centimeter- and a meter - sized object with a density of 2 g @xmath33 for four values of the gas temperature . as shown here , increasing the temperature of the gas results in increasing the rate of radial migration .
this is an expected result which is the consequence of an increase in the radial component of the pressure gradients of the gas .
figure 8 shows this concept in more detail . in this figure , we have plotted the density of the gas as a function of @xmath3 on a plane parallel to the midplane at @xmath63 au . as a result of increasing the temperature of the gas from 50 k to 1000 k ,
the maximum value of the gas density on this plane is increased by three orders of magnitude .
such an increase implies an increase in the radial component of the pressure gradients of the gas which in turn results in more rapid radial migration . from equation ( 5 ) ,
the radial position of the maximum of the gas density at a certain height @xmath5 approaches smaller values by increasing the gas temperature .
figure 8 shows this for @xmath610.2 au . for an object at this height and at an initial radial position of
@xmath60 au , such a decrease in the radius of the local density enhancement causes the distance of the initial outward migration of the object to become smaller ( figure 7 ) .
for the temperature of 1000 k , the radial location of the maximum gas density at @xmath63 au becomes smaller than 2 au and as a result , an object at ( 2,0.2 ) au does not undergo an initial outward migration . while the combined effect of pressure gradients and the drag force of the gas causes solids to radially migrate toward the location of local density enhancement , the vertical component of the gravitational force of the central star attracts solids toward the midplane of the nebula .
figure 9 shows the vertical motion of solids with different sizes .
the dependence of the rate of vertical migration on the size of an object can clearly be seen from this figure . as shown here , except for when the objects are in the vicinity of the midplane , the rates of their vertical migrations increase by increasing their radii . in the vicinity of the midplane , however , while 10 centimeter - sized and smaller objects continue their smooth descent , the 1 meter - sized object undergoes an overshoot .
such an overshoot and its corresponding damped oscillatory motion are more pronounced when the density of the object is increased .
figure 10 shows the vertical migrations of a 1 centimeter - sized and a 1 meter - sized object for different values of their densities . as shown here , the rates of vertical descent of both objects increase by increasing their densities .
once in the vicinity of the midplane , the meter - sized object undergoes a damped oscillatory motion whose amplitude increases by increasing the solid s density .
figure 11 shows the three - dimensional paths of these two objects .
the dynamical behavior of an object along the @xmath5-axis and in the vicinity of the midplane can be explained by studying equation ( 18 ) in more detail . in this equation ,
the drag coefficient @xmath64 is the factor that determines the functional form of @xmath65 . substituting for the drag force of the gas from equation ( 8) and for the @xmath5-component of the relative velocity of an object with respect to the gas from equation ( 12 ) , equation ( 18 )
can be written as @xmath66 { { \hat p}_z}\,.\ ] ] in the following we study equation ( 20 ) for @xmath67 and @xmath68 .
the case of @xmath69 is not studied here since it corresponds to the motions of objects with sizes equal or greater than 100 m. as shown in hb03 , the rates of migrations of such objects due to gas drag and pressure gradient decrease by increasing their sizes . in this paper , we focus our study on rapid migration of small solids . replacing @xmath71 and @xmath72 by their numerical values and considering spherical solids with masses of @xmath73 , equation ( 20 ) is written as @xmath74}\ , { t^{1/2}}\,{{\hat p}_z},\ ] ] where @xmath15 is in kelvin . in the vicinity of the midplane , @xmath75 and equation ( 21 ) can be simplified to @xmath76}\ , { t^{1/2}}\,{\dot{\hat z}}+ { { \hat z}\over { { \hat r}^3 } } \simeq 0.\ ] ] equation ( 22 ) resembles the equation of a damped harmonic motion with a dimensionless frequency of @xmath77\bigr\}^{-2}}\,.\ ] ] equation ( 23 ) indicates that @xmath78 is a function of the solid s radius and density , the gas temperature and also the mean free path of the gas molecules , @xmath31 .
we recall that @xmath31 varies with the gas density as mentioned in @xmath03 . at any location in the vicinity of the midplane ,
the type of the motion of an object depends on the sign of @xmath78 at that location .
the left column of figure 12 shows the absolute value of the second term of equation ( 23 ) for a centimeter - sized object initially at ( 1.25,0.125 ) au .
we chose @xmath79 cm as a representative of solids with @xmath67 since the results of our numerical integrations indicate that the first sign of a reynolds number larger than unity appears for an object with a radius of 10 cm . since for an object at ( 1.25,0.125 )
au , @xmath80 varies between 0.512 and 1 , the graphs of @xmath81 , as shown in figure 12 , indicate that for a centimeter - sized object , @xmath78 is negative .
a negative value for @xmath78 is an indication of an overdamped motion .
increasing the density of the solid while keeping the gas temperature constant , @xmath78 becomes larger and approaches zero ( critically damped ) indicating that the overdamped nature of the vertical motion of the particle weakens and it may even change into an underdamped motion where @xmath78 is positive . from equation ( 23 ) , the overdamping nature of the vertical motion becomes even more pronounced when the radius of the object is smaller than 1 cm .
that means , the rate of vertical migration of a solid decreases by decreasing its size ( figure 9 ) .
the left column of figure 12 also shows that for a constant value of the solid s density , increasing the gas temperature results in lower values for @xmath78 .
that means , at higher temperatures , the overdamping of the vertical motion of the object becomes more pronounced ( top graph of figure 13 ) .
this can be explained noting that from equation ( 5 ) and as shown in figure 15 , at any radial distance @xmath3 , the vertical distribution of the gas maximizes at @xmath82 ( i.e. , on the midplane ) . as shown in figure 14 , the width of the gas distribution function along the @xmath5-axis increases by increasing the gas temperature . as a result of this broadening ,
the resistive effect of gas drag on the vertical motion of solids appear in a larger range of variable @xmath5 which results in a longer time for descending toward the midplane .
in this case , equation ( 20 ) can be written as @xmath84}\ , { t^{1/2}}\,{{\hat p}_z } \nonumber\\ & \biggl\{1 + 8.56\,{[{a_p}({\rm { cm}})]^{0.4}}\,{t^{-0.2 } } { \biggl[{{\hat p}_r^2}+{{\hat p}_z^2}+ \biggl({{{\hat p}_\varphi}\over { \hat r } } - { \sqrt { { 1\over { \hat r } } - { 9.29 \times 10^{-6}}\,t\ , { \hat r}({\hat r}-1)}\,\biggr)^2}\,\biggr]^{0.2}}\nonumber\\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad { \rm { exp}}\bigl\{86098\>{t^{-1 } } \biggl[{1\over { { ( { { \hat r}^2}+{{\hat z}^2})^{1/2}}}}- { 1\over { \hat r}}\biggr]-0.4{({\hat r}-1)^2}\bigr\}\biggr\}.\end{aligned}\ ] ] similar to the previous case , in the vicinity of the midplane , equation ( 24 ) can be simplified to @xmath85}\ , { t^{1/2}}\,{\dot{\hat z } } \biggl\{1 + 8.56\ , { [ { a_p}({\rm { cm}})]^{0.4}}\,{t^{-0.2}}\nonumber\\ & \qquad { \biggl[{{\dot{\hat r}}^2}+{{\dot{\hat z}}^2}+ \biggl({\hat
r}{\dot{\hat \varphi } } - { \sqrt { { 1\over { \hat r } } - { 9.29 \times 10^{-6}}\,t\ , { \hat r}({\hat r}-1)}\,\biggr)^2}\,\biggr]^{0.2 } } { e^{-0.4{({\hat r}-1)^2}}}\biggr\}+ { { \hat z}\over { { \hat r}^3 } } \simeq 0.\end{aligned}\ ] ] for any value of @xmath86 , equation ( 25 ) resembles the equation of motion of a damped oscillator with a dimensionless frequency of @xmath87\bigr\}^{-2}}\nonumber\\ & \!\!\!\!\!\ ! { \biggl\{1 + 8.56 \,{[{a_p}({\rm { cm}})]^{0.4}}\,{t^{-0.2 } } { \biggl[{{\dot{\hat r}}^2}\!+\!{{\dot{\hat z}}^2}\!+\ ! \biggl({\hat r}{\dot{\hat \varphi}}\!-\ ! { \sqrt { { 1\over { \hat r } } - { 9.29 \times 10^{-6 } } t { \hat r}({\hat r}-1)}\,\biggr)^2}\,\biggr]^{0.2}}\!\!\!\ ! { e^{-0.4{({\hat r}-1)^2}}}\biggr\}^2}\!\!\!.\end{aligned}\ ] ] the right column of figure 12 shows the absolute value of the second term of this equation for different values of the density of a meter - sized object initially at ( 1.25,0.125 ) au .
the radial migration of such an object starts at @xmath88 and terminates at @xmath89 , implying that the first term of equation ( 26 ) will vary between 0.512 and 1 .
as shown in figure 12 , at a constant temperature and for a given value of @xmath86 , the contribution of the second term of equation ( 26 ) is quite small . as a result
, the instantaneous dimensionless frequency @xmath78 stays positive for the entire time of the solid s migration indicating that the motion of such an object , in both sides of the midplane , is an underdamped oscillatory motion .
figure 12 also shows that the numerical contribution of the second term of equation ( 26 ) decreases by increasing the solid s density at a constant temperature .
this results in larger frequencies for denser objects which in turn , results in more oscillations ( figure 10 ) .
one can also see from figure 12 that for a constant value of the solid s density , increasing the gas temperature will increase the numerical value of the second term of equation ( 26 ) .
this results in a smaller frequency and less oscillatory behavior ( bottom graph of figure 13 ) .
it is necessary to mention that the irregular shape of the second term of equation ( 26 ) for @xmath90 , is due to the initial outward migration of the solid .
we have studied the migration of solids induced by gas drag and pressure gradients in a non - uniform nebula .
we considered a rotating , isothermal and turbulence - free gaseous disk with an azimuthally symmetric density enhancement on its midplane and a solar type star at its center .
the results of our numerical simulations of the motions of solids ranging from micron - sized particles to meter - sized objects indicated that solids tend to migrate toward the locations of local maximum pressure . in our model
nebula , such locations correspond to places where the density of the gas is locally enhanced .
our results indicate that depending on the physical properties of solids , gas drag induced migration can be quite rapid .
such rapid migrations can increase the rates of collision and coalescence of particles and enhance their growth - rates to larger bodies .
the results of the extension of this study to include mass - growth and collisional coagulation are currently under preparation for publication .
it is necessary to emphasize that the assumptions of an isothermal and turbulence - free nebula were made merely to focus attention on the sole effect of gas drag on the rate of migration of solids .
also , in this study , we focused our attention on the motion of objects without considering their mutual interactions . in a realistic scenario
, one has to allow for the changes of the gas temperature on both radial and vertical directions and also take turbulence and collision into consideration .
it is expected that by including turbulence , in particular in the vertical direction , the rate of approach of particles to the midplane of the nebula may decrease .
it would then be of utmost value to investigate to what degree the combined effect of turbulence and gas drag induced rapid migration will enhance the rate of the formation of larger bodies .
such studies are currently underway .
adachi , i. , hayashi , c. , @xmath91 nakazawa , k. 1976 , prog.theor.phys .
, 56 , 1756 haghighipour , n. @xmath91 boss , a. p. 2003 , to appear in supulver , k. d. , @xmath91 lin , d. n. c. 2000 , icarus , 146 , 525 takeuchi , t. , @xmath91 lin , d. n. c. 2002 , , 581 , 1344 weidenschilling , s. j. 1977 , , 180 , 57 whipple , f. l. 1964 , proc . | we study the motions of small solids , ranging from micron - sized dust grains to meter - sized objects , in the vicinity of local pressure enhancements of a gaseous nebula .
integrating numerically , we show that as a result of the combined effect of gas drag and pressure gradients , solids tend to accumulate at the locations where the pressure of the gas maximizes .
the rate of migration of solids varies with their sizes and densities and also with the physical properties of the gas .
the results of our numerical simulations indicate that such migrations are most rapid for meter - sized objects .
the applicability of the results to the enhancement of the collision and coagulation of solids and also to the growth - rate of planetesimals is discussed . |
the existence of exotic states beyond the constituent quark model ( cqm ) has been speculated about almost since the introduction of colour @xcite . due to the self - coupling of gluons via colour - charge ,
so - called hybrid mesons and glueball are allowed within quantum chromodynamics while they are beyond the cqm .
hybrid mesons are @xmath4 states with an admixture of gluons , and glueballs are states with no quark content , only consisting of ( constituent ) gluons . according to lattice - qcd predictions @xcite , the glueball candidate lowest in mass has scalar quantum numbers , @xmath5 , and is expected to have a mass of @xmath6 .
a glueball candidate has been observed by the crystal barrel and the wa102 experiments , however , mixing with ordinary isoscalar mesons makes the interpretation difficult .
several light hybrids , on the other hand , are predicted to have exotic @xmath0 quantum numbers and are thus promising candidates in the search for physics beyond the cqm .
the lowest mass hybrid candidate for example is predicted @xcite to have exotic quantum numbers of spin , parity and charge conjugation @xmath7 not attainable by ordinary @xmath4 states , and a mass between 1.3 and 2.2gev / c@xmath8 .
two experimentally observed @xmath9 hybrid candidates in the light - quark sector have been reported in the past in different decay channels , the @xmath10 mainly seen in @xmath11 decays , by e.g. e852 @xcite , ves@xcite , and crystal barrel @xcite , and the @xmath3 , observed by both e852 and ves in the decay channels : @xmath12 @xcite , @xmath13 @xcite , @xmath14 @xcite , and @xmath15 @xcite .
in particular the resonant nature of the @xmath12 decay channel of the @xmath3 observed in @xmath16 final states is highly disputed @xcite .
compass has started to shed new light on the puzzle of spin - exotics by the observation of an @xmath9 signal in the 2004 data , consistent with the famous @xmath3 .
it shows clean phase motions with respect to other waves , confirming the resonance nature @xcite .
the compass two - stage spectrometer @xcite at the cern sps features electromagnetic calorimetry in both stages .
photon detection in a wide angular range at high resolution is crucial for decay channels involving @xmath17 , @xmath2 or @xmath18 , the ring imaging cherenkov ( rich ) detector in the first stage allows for final state particle identification ( pid ) .
a good separation of pions from kaons enables the study of kaonic final states .
two cherenkov differential counters with acromatic ring focus ( cedar ) upstream of the target are used to identify the incoming beam particle .
not only production of strangeness with the pion beam can thus be studied but also kaon diffraction , tagging the kaon contribution in the negative hadron beam ( 96% @xmath19 , 3.5% @xmath20 , 0.5% @xmath21 ) . after a short pilot run in 2004 of 190
gev / c @xmath19 beam on a lead target ( sec.[subsec.2004 ] ) , compass recorded data with 190gev / c hadron beams in 2008/09 , providing excellent opportunity for simultaneous observation of new states in various decay modes within the same experiment ( sec.[subsec.2008 ] , [ subsec.2008b ] ) .
moreover , the data contain subsets with different beam projectiles ( @xmath22 ) and targets ( h@xmath23 , ni , w , and pb ) , allowing for systematic studies not only of diffractive and central production but also primakoff reactions @xcite and baryon spectroscopy @xcite .
the quantum numbers spin @xmath25 , parity @xmath26 and @xmath27-parity of the produced resonance @xmath28 , together with the spin projection given by @xmath29 and @xmath30 ( reflectivity ) , define a partial wave @xmath31l$ ] . the partial wave analysis ( pwa )
is based on the isobar model , see fig.[fig : diffrprod_spectro ] _ ( right)_. the resonance @xmath28 decays via an intermediate di - pion resonance ( the isobar ) , accompanied by a so - called bachelor pion .
the pwa method consists of two steps .
first , a mass - independent fit is performed on the data binned into 40mev / c@xmath8 wide mass intervals , no assumption on the resonance structure of the @xmath16 system is made at this level .
a total set of 42 waves including a flat background wave is fitted to the data using an extended maximum likelihood method , which comprises acceptance corrections .
subsequently , the mass - dependent fit is applied to the six main waves out of the result from the first step , and uses a @xmath32 minimisation .
the mass dependence is parameterised by relativistic breit - wigners ( bw ) and coherent background , if present .
the employed parameterisation of the spin density matrix has a rank of two , accounting for spin - flip and spin - non - flip amplitudes at the baryon vertex .
fig.[fig : pwa2004 ] shows the intensity of the @xmath33 wave with the well - established @xmath34 and that of the spin - exotic @xmath9 wave as well as the phase difference @xmath35 between the two , for the fits of all six waves , see @xcite .
the black data points represent the mass - independent fit , whereas the mass - independent one is overlayed as solid line , the separation into background ( dotted ) and bw ( dashed ) is plotted where applicable .
especially the resonant nature of the exotic @xmath36 p$ ] wave is questioned in previous observations , whereas our data shows a clear and rapid phase motion .
our result of a mass of @xmath37mev / c@xmath8 and a width of @xmath38mev / c@xmath8 is consistent with the famous @xmath3 @xcite already reported in the past but still controversially discussed .
an important cross - check of all analyses is the test for isospin symmetry in the observed spectra .
the @xmath12 decay channel of the @xmath3 for example , can be studied in two modes of 3@xmath39 final states , @xmath40 ( charged ) and @xmath41 ( neutral ) , respectively .
the relative contribution should follow isospin conservation , depending on the underlying isobars , as it is shown in fig.[fig : pwa2008 ] . a first partial - wave analysis ( pwa ) of main waves in diffractively produced 3@xmath39 events
has been performed for both modes , applying the same model as for the 2004 result , for details see @xcite . the wave intensities shown are normalised in the same way to the well - known @xmath12 decay of the @xmath42 to compensate for the different statistics analysed , thus making them comparable .
we find similar intensities for the @xmath12 decay , whereas a suppression factor of two is observed for the wave decaying into @xmath43 as expected due to the clebsch - gordon coefficients .
further ongoing analyses of neutral channels cover @xmath44 and @xmath45 final states ( search for the @xmath10 and lightest @xmath46 glueball candidate ) as well as @xmath47 , @xmath48 and @xmath49 final states ( accessible isobars : @xmath50 ) .
for all these channels , compass has recorded significantly higher statistics with respect to previous experiments , covering all spin - exotic meson decay channels in the light quark sector reported in the past .
final states including strange particles are interesting for both , glueball search in central production as well as diffractively produced hybrids .
fig.[fig : kaons ] shows the @xmath51 subsystems out of the @xmath52 system , again for two different modes : @xmath53 ( charged mode ) and @xmath54 ( neutral mode ) final states , respectively . in both cases ,
the spectra show a clear structure around the expected @xmath55 .
the cedars were used to anti - tag the kaons in the beam , and the final state kaons are identified using the rich detector and the well resolved @xmath56 secondary vertex , respectively ; for details on the event selections , see @xcite .
further ongoing analyses of kaonic final states cover the complementary @xmath57 system diffractively produced with the @xmath19 beam as well as kaon diffraction into @xmath58 final states using the beam kaons . in both cases ,
the rich detector is used for final state pid and the cedars for identifying the beam particle .
fig.[fig : kaons_b ] shows mass spectra for the study of kaon diffraction into @xmath58 final states . the total mass spectrum and the @xmath59 show the prominent , well - known resonances as expected , similar as observed by wa03 .
our statistics collected in merely 2008 exceeds the one from wa03 by a factor of two , for details , see @xcite and references therein .
the pwa of all kaonic final states mentioned in this paper are under preparation .
compass has taken data with high - intensity , negatively as well as positively charged hadron beams ( @xmath22 ) on nuclear and liquid hydrogen targets .
the data sample newly taken in 2008/09 exceeds the world data by a factor of 10 - 100 , depending on the given final state .
the compass data sets allow to address open issues in light - mesons spectroscopy at good accuracy , even extending the region to higher masses beyond 2gev / c@xmath8 .
this work is supported by the bmbf ( germany ) , particularly the `` nutzungsinitiative cern '' .
9 r. jaffe and k. johnsons , _ phys .
b _ * 60 * ( 1976 ) 201 .
t. barnes _ et .
_ , _ nucl .
b _ * 224 * ( 1983 ) 241 . c. mcneile and c. michael , _ phys . rev . _ * d73 * ( 2006 ) 074506 .
juge , j. kuti , c. morningstar , _ aip conf .
* 688 * ( 2004 ) 193 .
d. r. thomson _ et al .
_ , _ phys .
* 79 * ( 1997 ) 1630 .
g. m. beladidze _ et al .
* b313 * ( 1993 ) 276 .
a. abele _ et al .
_ , _ phys .
lett . _ * b423 * ( 1998 ) 175 .
g. s. adams _ et al .
_ , _ phys . rev
. lett . _ * 81 * , ( 1998 ) 5760 .
y. khokhlov , _ nucl .
phys . _ * a663 * ( 2000 ) 596 .
e. i. ivanov _ et al .
_ , _ phys .
rev lett . _
* 86 * ( 2001 ) 3977 .
j. kuhn _
_ , _ phys .
lett . _ * b595 * ( 2004 ) 109 .
d. v. amelin _ et al .
_ , _ phys . atom .
_ * 68 * ( 2005 ) 359 .
m. lu _ et al .
_ , _ phys .
lett . _ * 94 * ( 2005 ) 032002 .
dzierba _ et al .
_ , _ phys .
* 73 * ( 2006 ) 072001 .
m. alekseev _ et al .
_ , compass collaboration , _ phys .
* 104 * ( 2010 ) 241803 .
p. abbon _ et al .
_ , compass collaboration , _ nucl .
instrum . meth .
a _ * 577 * ( 2007 ) 455 . s. grabmller , _
meson2010 conf .
proc . _ submitted to _ int .
j. of modern phys . _ * a * ( 2010 ) .
a. austregesilo , these proceedings , _
inpc2010 , vancouver _ ( 2010 ) .
f. nerling , _ aip conf .
proc . _ * 1257 * ( 2010 ) 286 ; arxiv:1007.2951[hep - ex ] .
t. schlter , _ aip conf .
proc . _ * 1257 * ( 2010 ) 462 .
p. jasinski , _ spin - praha-2009 conf .
proc . _ submitted to _ eur .
j. _ * c * ( 2009 ) . | the compass fixed - target experiment at cern sps is dedicated to the study of hadron structure and dynamics . in the physics programme using hadron beams , the focus is on the detection of new states , in particular the search for @xmath0 exotic states and glueballs . after a short pilot run in 2004 ( 190 gev / c negative pion beam , lead target )
, we started our hadron spectroscopy programme in 2008 by collecting an unprecedented statistics with a negative hadron beam ( 190 gev / c ) on a liquid hydrogen target .
a similar amount of data with positive hadron beam ( 190 gev / c ) has been taken in 2009 , as well as some additional data with negative beam on nuclear targets .
the spectrometer features a large angular acceptance and high momentum resolution and also good coverage by electromagnetic calorimetry , crucial for the detection of final states involving @xmath1 or @xmath2 .
a first important result is the observation of a significant @xmath0 spin exotic signal consistent with the disputed @xmath3 in the pilot run data .
this result was recently published .
we present an overview of the status of various ongoing analyses on the 2008/09 data . |
in this paper , we continue our previous work @xcite by extending its results to the critical galton
watson trees whose offspring distribution may have infinite variance . to be more precise ,
let @xmath9 be a non - degenerate probability measure on @xmath10 with mean one , and we assume throughout this paper that @xmath9 is in the domain of attraction of a stable distribution of index @xmath7 $ ] , which means that @xmath11 where the function @xmath12 is slowly varying as @xmath13 .
the finite variance condition for @xmath9 is sufficient for the previous statement to hold with @xmath14 .
when @xmath15 , by classical results of ( * ? ?
* chapters xiii and xvii ) , the condition ( [ eq : stable - attraction ] ) is satisfied if and only if the tail probability @xmath16 varies regularly with exponent @xmath17 as @xmath18 . under the probability measure @xmath19 , for every integer @xmath20 , we let @xmath21 be a galton watson tree with offspring distribution @xmath9 , conditioned on non - extinction at generation @xmath2 . conditionally given the tree @xmath21 , we consider a simple random walk on @xmath21 starting from the root .
the probability distribution of its first hitting point of generation @xmath2 will be called the harmonic measure @xmath22 , which is supported on the set @xmath23 consisting of all vertices of @xmath21 at generation @xmath2 .
let @xmath24 be the probability that a critical galton
watson tree @xmath25 survives up to generation @xmath2 .
it has been shown by slack @xcite that , as @xmath26 , the probability @xmath27 decreases as @xmath28 up to multiplication by a slowly varying function , and @xmath29 converges in distribution to a non - trivial limit distribution on @xmath30 that we will specify shortly .
the main result of the present work generalizes theorem 1 of @xcite from the finite variance case ( with @xmath14 ) to all @xmath31 $ ] .
[ thm : dim - discrete ] let @xmath32 be a random vertex uniformly chosen from @xmath33 . if the offspring distribution @xmath9 has mean one and belongs to the domain of attraction of a stable distribution of index @xmath0 $ ] , there exists a constant @xmath34 , which only depends on @xmath5 , such that for every @xmath35 , we have @xmath36 roughly speaking , the previous theorem asserts that if we look at a typical vertex at level @xmath2 of the conditional galton
watson tree @xmath21 , then this random vertex carries with high probability a mass of order @xmath37 given by the harmonic measure @xmath1 . since @xmath38
, we see particularly that the harmonic measure on the conditional galton
watson tree is not uniformly spread and exhibits a fractal behavior . the fractal nature of the harmonic measure in critical galton
watson tree has been initially investigated by curien and le gall @xcite from another perspective . to compare with theorem [ thm : dim - discrete ] ,
their main result in the finite variance case has also been generalized to the present setting as theorem 1.1 in @xcite : under the same assumption of theorem [ thm : dim - discrete ] , we have the existence of another constant @xmath39 only depending on @xmath5 , such that for every @xmath35 , we have the convergence in @xmath40-probability @xmath41{(\mathbb{p } ) } 1\ , .
\label{eq : theorem1-lin1}\ ] ] in other words , with high probability the mass given by the harmonic measure @xmath1 to a random vertex at level @xmath2 drawn with respect to the same harmonic measure is of order @xmath42 with @xmath43 .
we can thus say that the harmonic measure @xmath1 is mainly supported on a subset of size approximately equal to @xmath44 , which is much smaller than the whole size of @xmath33 .
notice that the family @xmath45 is shown in ( * ? ? ?
* theorem 1.3 ) to be bounded from above in @xmath46 .
as the hitting distribution @xmath22 of generation @xmath2 by simple random walk on @xmath21 is unaffected if we remove the branches of @xmath21 that do not reach height @xmath2 , we can simplify the model by considering merely simple random walk on the reduced tree @xmath47 associated with @xmath21 , which consists of all vertices of @xmath21 that have at least one descendant at generation @xmath2 .
when the critical offspring distribution @xmath9 has infinite variance , the study of scaling limits of @xmath47 goes back to vatutin @xcite and yakymiv @xcite . if we scale the graph distance by the factor @xmath48 , as @xmath26 the discrete reduced tree @xmath49 converges to a random continuous tree @xmath50 that we now describe . for every @xmath7 $ ] , we define the @xmath5-offspring distribution @xmath51 as follows . if @xmath14 , we let @xmath52 be the dirac measure at 2 . if @xmath53 , @xmath51 is the probability measure on @xmath54 given by @xmath55 where @xmath56 denotes the usual gamma function of euler .
the mean of @xmath51 is equal to @xmath57 . to construct the random tree @xmath58 , we begin with the genealogical tree @xmath59 of a family of particles that evolves in continuous time according to the following rules . at time 0
there is a single particle , this particle lives for a random time uniformly distributed over @xmath60 $ ] , then dies and gives birth to a random number of new particles .
this number of offspring is distributed according to @xmath61 and is assumed to be independent of the lifetime of the initial particle .
inductively , all the particles evolve independently of each other , and each new particle appeared at time @xmath62 dies at a time uniformly distributed over @xmath63 $ ] and gives birth to an independent number of new particles according to the supercritical offspring distribution @xmath61 . after taking completion of @xmath59 with respect to its natural intrinsic metric @xmath64 , we obtain a random compact rooted @xmath65-tree @xmath50 that will be called the reduced stable tree of parameter @xmath5 .
its boundary @xmath66 is composed of all points of @xmath58 at height 1 .
we refer to section [ sec : treedelta ] for a precise definition of @xmath58 .
the description above makes clear the recursive structure of @xmath58 : let @xmath67 be a uniform random variable over @xmath60 $ ] and let @xmath68 be a random variable of law @xmath61 .
we take @xmath69 to be a sequence of independent copies of @xmath58 , and we assume the independence between @xmath70 and @xmath71 .
if we attach to the top of a single line segment of length @xmath67 the roots of the rescaled trees @xmath72 , the resulting tree rooted at the origin of the initial line segment will have the same distribution as @xmath73 .
see fig .
[ fig : recur ] for an illustration .
recursive structure of the reduced stable tree @xmath58,width=340 ] as the continuous analog of simple random walk , brownian motion on @xmath58 starting from the root can be defined up until its first hitting time of @xmath74 .
it behaves like linear brownian motion as long as it stays inside a line segment of @xmath58 .
it is reflected at the root of @xmath58 and when it arrives at a branching point , it chooses one of the adjacent line segments with equal probabilities .
we define the ( continuous ) harmonic measure @xmath75 on @xmath58 as the ( quenched ) distribution of the first hitting point of @xmath74 by brownian motion .
the behavior of @xmath75 is closely related to that of the discrete harmonic measure @xmath1 . in order to state a result
analogous with theorem [ thm : dim - discrete ] in the continuous setting , we must first make sense of a `` typical '' point chosen from the boundary @xmath74 . to this end
, we introduce another ( non - compact ) random rooted @xmath65-tree @xmath76 endowed with the same branching structure as @xmath58 , such that each point of @xmath76 at height @xmath77 corresponds bijectively to a point of @xmath59 at height @xmath78 .
it is easy to verify that the resulting new tree @xmath76 is a continuous - time galton
watson tree of branching rate 1 with the supercritical offspring distribution @xmath61 .
in particular , @xmath79 is the yule tree which describes the genealogy of the classical yule process . by definition ,
the boundary @xmath80 of @xmath76 is the set of all infinite geodesics in @xmath76 starting from the root ( these will be called geodesic rays ) . since @xmath58 and @xmath76 share the same branching structure , both @xmath74 and @xmath80 can be canonically identified with a common random subset of @xmath81 . for every @xmath82 , we write @xmath83 for the level set of @xmath76 at height @xmath84 , and we know that @xmath85= \exp ( \frac{r}{\alpha-1})$ ] . by a martingale argument
, one can define @xmath86 as @xmath87 , the kesten
stigum theorem ( for continuous - time galton watson trees , see e.g. ( * ? ? ?
* theorem iii.7.2 ) ) implies that the previous convergence holds in the @xmath88-sense , @xmath89=1 $ ] , and @xmath90 almost surely .
it follows from theorem iii.8.3 in @xcite that @xmath91 = 1-\frac{u}{(1+u^{\alpha-1})^{\frac{1}{\alpha-1 } } } \quad \mbox { for any } u\in ( 0,\infty).\ ] ] moreover , according to theorem 1 of @xcite , @xmath29 converges in distribution to this martingale limit @xmath92 as @xmath26 . for every @xmath93 , we let @xmath94 denote the height of @xmath95 in @xmath76 , and we write @xmath96 $ ] for the tree of descendants of @xmath95 in @xmath76 , viewed as an infinite random @xmath65-tree rooted at @xmath95 . for every @xmath82 , we write @xmath97 $ ] for the level set at height @xmath84 of the tree @xmath96 $ ] .
if one thinks of @xmath96 $ ] as a subtree of @xmath76 , the set @xmath97 $ ] consists of all the points of @xmath76 at height @xmath98 that are descendants of @xmath95 .
we similarly define @xmath99 \colonequals \lim_{r\to \infty } e^{-\frac{r}{\al-1}}\ # \gamma\a_r[x],\ ] ] which has the same distribution as @xmath100 .
the uniform measure @xmath101 on @xmath80 is defined as the unique probability measure on @xmath80 satisfying that , for every @xmath93 and for every geodesic ray @xmath102 passing through @xmath95 , @xmath103}{\mathcal{w}\a},\ ] ] where @xmath104 stands for the set of all geodesic rays in @xmath76 that coincide with @xmath105 up to height @xmath94 . in existing literature , we also call @xmath101 the branching measure on the boundary of @xmath76 .
since @xmath74 can be identified with @xmath80 as explained above , we let @xmath106 be the ( random ) probability measure on @xmath74 induced by @xmath101 , which will be referred to as the uniform measure on @xmath74 .
the following result is an extension of theorem 2 in @xcite .
[ thm : dim - continu ] for every @xmath107 $ ] , with the same constant @xmath108 as in theorem [ thm : dim - discrete ] , we have @xmath19-a.s . @xmath109-a.e .
@xmath110 where @xmath111 stands for the closed ball of radius @xmath84 centered at @xmath105 in the metric space @xmath112 . by duquesne and le gall (
* theorem 5.5 ) , the hausdorff measure of @xmath74 with respect to @xmath64 is @xmath19-a.s .
equal to @xmath113 .
according to lemma 4.1 in @xcite , assertion of the preceding theorem implies that @xmath19-a.s .
the uniform measure @xmath106 has the same hausdorff dimension as the whole boundary of the reduced stable tree .
meanwhile , taking account of , we may interpret @xmath108 as the local dimension of the harmonic measure @xmath114 at a typical point of the boundary @xmath74 .
in contrast , the hausdorff dimension of the harmonic measure @xmath114 is @xmath19-a.s .
equal to the constant @xmath115 appearing in ( [ eq : theorem1-lin1 ] ) , since theorem 1.2 of @xcite shows that @xmath19-a.s .
, @xmath117 since @xmath118 , @xmath19-a.s .
the set @xmath119 satisfies that @xmath120 whereas @xmath121 .
[ corol - intro ] for every @xmath107 $ ] , @xmath19-a.s .
the two measures @xmath114 and @xmath106 on the boundary of @xmath58 are mutually singular . when @xmath5 approaches 1 , the hausdorff dimension @xmath115 of the harmonic measure @xmath114 remains bounded ( see theorem 1.3 in @xcite ) .
naively , it would seem that the typical local dimension @xmath122 of @xmath114 would behave asymptotically like @xmath123 for some positive constant @xmath124 .
however , the explosion of @xmath122 turns out to be much faster than @xmath125 when @xmath5 decreases to 1 .
[ prop : dimension - explosion ] we have that @xmath126 our proof of proposition [ prop : dimension - explosion ] relies on the fact that the constant @xmath122 appearing in theorems [ thm : dim - discrete ] and [ thm : dim - continu ] can be expressed explicitly with help of the conductance of @xmath58 .
informally , if we think of the random tree @xmath58 as an electric network of resistors with unit resistance per unit length , the effective conductance @xmath127 between the root and the boundary @xmath74 is a random variable larger than 1 . from a probabilistic point of view
, it is the mass under the brownian excursion measure for the excursion paths away from the root that hit height 1 . following the definition of @xmath58 and its electric network interpretation
, the distribution of @xmath127 satisfies the recursive distributional equation @xmath128 where @xmath129 are i.i.d .
copies of @xmath127 , the integer - valued random variable @xmath130 is distributed according to @xmath51 , and @xmath67 is uniformly distributed over @xmath60 $ ] .
all these random variables are supposed to be independent .
[ prop : dim - formula ] for any @xmath31 $ ] , the constant @xmath122 appearing in theorems [ thm : dim - discrete ] and [ thm : dim - continu ] is given by @xmath131 - 1.\ ] ] moreover , @xmath122 is decreasing for all @xmath107 $ ] .
the surprisingly simple formula ( [ eq : lal - value ] ) provides a unified expression of @xmath122 in terms of @xmath100 and @xmath132 .
we will see in section [ sec : conductance ] that the product @xmath133 has the same first moment as the conductance @xmath134 of a size - biased version of the reduced stable tree @xmath58 .
note that for @xmath135 , the @xmath136-offspring distribution @xmath61 is decreasing for the usual stochastic partial order .
this property allows one to construct simultaneously all reduced stable trees @xmath137 $ ] as a nested family , so that @xmath138 for all @xmath139 ( see section 2.4 in @xcite ) .
while the monotonicity of the typical local dimension @xmath122 is affirmed by the previous result , one question still open is whether the hausdorff dimension @xmath115 of the continuous harmonic measure @xmath114 is also decreasing with respect to @xmath136 .
the rest of this paper is divided into three parts .
the continuous model of brownian motion on @xmath58 is studied in section [ sec : continuous ] , where we prove theorem [ thm : dim - continu ] , propositions [ prop : dimension - explosion ] and [ prop : dim - formula ] .
then we set up notation and terminology for the discrete setting in section [ sec : discrete ] , while section [ sec : proof - thm1 ] is devoted to proving theorem [ thm : dim - discrete ] .
the arguments are basically adapted from those used in the finite variance case .
however , we emphasize that several modifications are indispensable to circumvent the problem of infinite variance . instead of repeating some highly analogous reasoning , we refer the reader to @xcite for details that we omit .
the author is grateful to an anonymous referee for many comments that greatly improved the paper .
let us begin with a formal definition of the reduced stable tree @xmath140 of parameter @xmath141 $ ] .
we set @xmath142 where by convention @xmath143 and @xmath144 .
if @xmath145 , we set @xmath146 the generation of the vertex @xmath147 ( in particular , @xmath148 ) , and if @xmath149 , we define the parent of @xmath147 as @xmath150 and then say that @xmath147 is a child of @xmath151 . for two elements @xmath152 and @xmath153 belonging to @xmath154 , their concatenation is @xmath155 .
the notions of a descendant and an ancestor of an element of @xmath156 are defined in the obvious way , with the convention that every @xmath157 is both an ancestor and a descendant of itself .
an infinite subset @xmath158 of @xmath154 is an infinite discrete ( rooted ordered ) tree without leaves if there exists a collection of positive integers @xmath159 for every @xmath160 such that @xmath161 recall that the generating function of the @xmath5-offspring distribution @xmath51 is given ( see for instance @xcite ) as @xmath162.\ ] ] for fixed @xmath0 $ ] , we introduce a collection @xmath163 of independent random variables distributed according to @xmath51 under the probability measure @xmath19 , and define a random infinite discrete tree @xmath164 we point out that @xmath165 is just a supercritical galton watson tree with offspring distribution @xmath61 .
in particular , @xmath166 is an infinite binary tree .
let @xmath167 be another collection , independent of @xmath163 , consisting of independent real random variables uniformly distributed over @xmath60 $ ] under the same probability measure @xmath19 .
we set now @xmath168 and then by induction , for every @xmath169 different from the root , @xmath170 .
note that a.s .
@xmath171 for every @xmath169 .
consider the set @xmath172 \big)\cup \bigg(\bigcup_{v\in\pi^{(\alpha)}\backslash\{\varnothing\ } } \{v\ } \times ( y_{\bar v } , y_v]\bigg).\ ] ] there is a straightforward way to define a metric @xmath173 on @xmath174 , so that @xmath175 is a ( non - compact ) @xmath65-tree and , for every @xmath176 , we have @xmath177 .
see [ fig : delta ] for an illustration of the tree @xmath174 when @xmath178 .
the random tree @xmath174 when @xmath179,width=453 ] the reduced stable tree @xmath58 is the completion of @xmath174 with respect to the metric @xmath173 .
it is immediate to see that @xmath180 is a compact @xmath65-tree , and @xmath181 where @xmath182 is the boundary of @xmath58 , which can be identified with a random subset of @xmath81 .
the point @xmath183 is called the root of @xmath50 . for every @xmath184
, we set its height @xmath185 .
a genealogical order on @xmath50 can be defined by setting @xmath186 if and only if @xmath95 belongs to the geodesic path from the root to @xmath187 .
conditionally on @xmath50 , we can define brownian motion @xmath188 on @xmath50 starting from the root and up to its first hitting time @xmath189 of @xmath66 .
see e.g. ( * ? ? ?
* section 2.1 ) for the details of this construction .
the harmonic measure @xmath75 is the distribution of @xmath190 , which is a ( random ) probability measure on @xmath191 . to introduce a new tree sharing the same branching structure as @xmath50 , we start with the same infinite discrete tree @xmath165 introduced in section [ sec : treedelta ] .
consider now a collection @xmath192 of independent real random variables exponentially distributed with mean @xmath193 under the probability measure @xmath19 .
we set @xmath194 and then by induction , for every @xmath169 different from the root , @xmath195 .
the continuous - time galton watson tree ( hereafter to be called ctgw tree for short ) of stable index @xmath5 is the set @xmath196\big ) \cup \bigg(\bigcup_{v\in \pi^{(\alpha)}\backslash\{\varnothing\ } } \{v\ } \times ( z_{\bar v } , z_v]\bigg),\ ] ] which is equipped with the metric @xmath197 defined in the same way as @xmath173 in the preceding subsection . for this metric
, @xmath198 is a.s .
an infinite @xmath65-tree . for every @xmath199
, we keep the notation @xmath200 for the height of @xmath95 .
observe that if @xmath67 is uniformly distributed over @xmath60 $ ] , the random variable @xmath201 is exponentially distributed with mean @xmath193 .
hence we may and will suppose that the collection @xmath192 is constructed from the collection @xmath202 in the previous subsection via the formula @xmath203 for every @xmath204 .
a homeomorphism from @xmath174 onto @xmath198 is given by the mapping @xmath205 defined as @xmath206 for every @xmath207 . by stochastic analysis
, we can write for every @xmath208 , @xmath209 where @xmath210 is brownian motion with constant drift @xmath211 towards infinity on the ctgw tree @xmath76 ( this process is defined in a similar way as brownian motion on @xmath50 , except that it behaves like brownian motion with drift @xmath211 on every line segment of the tree ) .
it is easy to see that the brownian motion @xmath212 is transient . from now on ,
when we speak about brownian motion on the ctgw tree or on other similar infinite trees , we will always mean brownian motion with drift @xmath211 towards infinity . by definition ,
the boundary of @xmath198 is the set of all geodesic rays in @xmath198 starting from the root @xmath183 , and it can be canonically embedded into @xmath81 . due to the transience of brownian motion on @xmath198 , there is an a.s .
unique geodesic ray denoted by @xmath213 that is visited by @xmath210 at arbitrarily large times .
the distribution of the exit ray @xmath213 yields a ( random ) probability measure @xmath214 on @xmath81 .
thanks to , we have in fact @xmath215 , provided we think of both @xmath75 and @xmath214 as probability measures on @xmath81 .
* infinite continuous trees .
* we let @xmath216 be the set of all pairs @xmath217 that satisfy the following conditions : 1 .
@xmath158 is an infinite discrete tree without leaves in the sense of section [ sec : treedelta ] ; 2 .
@xmath218 for all @xmath219 ; 3 .
@xmath220 for every @xmath221 ; 4 . for every @xmath222 , we have @xmath223 if @xmath224 , we consider the associated `` tree '' @xmath225\big ) \cup \bigg(\bigcup_{v\in\pi\backslash\{\varnothing\ } } \{v\ } \times ( z_{\bar v } , z_v]\bigg),\ ] ] equipped with the natural distance defined as above .
the set @xmath226 can be identified with the boundary @xmath227 of the tree @xmath228 , which is defined as the collection of all geodesic rays in @xmath228 .
we keep the notation @xmath229 for the height of a point @xmath230 .
the genealogical order on @xmath228 is defined as previously and again is denoted by @xmath231 . if @xmath232 and @xmath233 , we write @xmath234 if @xmath235 for some integer @xmath236 . when we say that we consider a tree @xmath237 , it means that we are given a pair @xmath217 satisfying the above properties , with @xmath228 being the associated tree .
clearly , for every @xmath31 $ ] , the ctgw tree @xmath198 is a random element in @xmath216 , and we write @xmath238 for its distribution .
let us fix @xmath239 . under our previous notation , the root @xmath240 has @xmath241 offspring .
we denote by @xmath242 the subtrees of @xmath228 rooted at the first branching point @xmath243 . to be more precise , for every @xmath244 , we define the shifted discrete tree @xmath245=\{v\in \mathcal{v}\colon iv\in \pi\}$ ] , and @xmath246 is the infinite continuous tree corresponding to the pair @xmath247,(z_{iv}-z_{\varnothing})_{v\in \pi[i]})$ ] . under @xmath238 ,
the offspring number @xmath241 is distributed according to @xmath51 .
the branching property of the ctgw tree means that under the conditional law @xmath248 , the subtrees @xmath249 are i.i.d .
following the same law @xmath250 . if @xmath82 , the level set of @xmath228 at height @xmath84 is @xmath251 .
if @xmath252 , let @xmath253 $ ] denote the subtree of descendants of @xmath95 in @xmath228 . to define it formally , we write @xmath254 for the unique element of @xmath156 such that @xmath255 , then @xmath253 $ ] is the tree associated with the pair @xmath256 , ( z_{v_xv}-r)_{v\in\pi[v_{x}]})$ ] . as usual , @xmath257 $ ] stands for the level set at height @xmath84 of the tree @xmath253 $ ] . as we have seen in the introduction , the martingale limit @xmath258 exists @xmath259-a.s . , and @xmath260
for every @xmath261 , we similarly set @xmath262)=\lim_{r\to \infty } e^{-\frac{r}{\al-1}}\ # \t_r[x].\ ] ] if @xmath263 is a geodesic ray passing through @xmath95 , let @xmath104 denote the set of geodesic rays in @xmath228 that coincide with @xmath105 up to height @xmath94 . then @xmath259-a.s .
, the uniform measure @xmath264 on @xmath227 is the probability measure on @xmath227 characterized by @xmath265)}{\mathcal{w}\a(\t)}\ ] ] for every @xmath266 such that @xmath234 . for a fixed infinite continuous tree @xmath228 , we define the harmonic measure @xmath267 on @xmath227 as the distribution of the exit ray chosen by brownian motion on @xmath228 ( with drift @xmath211 towards infinity ) .
we write @xmath268 for the set of all pairs consisting of a tree @xmath237 and a distinguished geodesic ray @xmath105 .
we define a transformation @xmath269 on @xmath270 by shifting @xmath271 at the first branching point of @xmath228 , that is , @xmath272 , where @xmath273 and @xmath274 is the tree rooted at the first branching point of @xmath228 that is chosen by @xmath105 .
the next result extends proposition 5 in @xcite from the yule tree @xmath79 to the ctgw tree @xmath76 .
[ prop : unif - meas - inv ] for all @xmath275 $ ] , the probability measure @xmath276 on @xmath270 is invariant under @xmath269 . fix a bounded measurable function @xmath277 on @xmath270 . under @xmath278 , the height @xmath279 of the first branching point is exponentially distributed with mean 1 , while the offspring number @xmath280 is distributed according to @xmath61 .
recall that @xmath281 and @xmath282 using these remarks and the branching property of the ctgw tree , we see that @xmath283 which shows the required invariance . to fully understand this invariant measure
, we construct a size - biased version @xmath284 of the ctgw tree @xmath76 , following the popularized idea of size - biasing a galton watson tree ( see e.g. @xcite ) .
let us start with the root @xmath240 .
it lives for a random exponential lifetime with parameter @xmath57 , then it dies and simultaneously gives birth to a random number @xmath285 of children , with @xmath285 distributed as the size - biased offspring number @xmath286 .
we pick one of these children uniformly at random , say @xmath287 . we give independently the other children independent descendant trees distributed as @xmath76 , whereas @xmath288 lives for another independent exponential lifetime with parameter @xmath57 and then reproduces a random number @xmath289 of children with @xmath289 an independent copy of @xmath285 . again
, we choose one of the children of @xmath288 uniformly at random , call it @xmath290 , and give the others independent ctgw descendant trees distributed as @xmath76 . repeating this procedure independently for infinitely many times
, we obtain a random infinite tree @xmath284 which will be called the size - biased ctgw tree with parameter @xmath107 $ ] .
a formal definition of @xmath284 with the distinguished geodesic ray @xmath291 as a random element of @xmath292 can be given similarly as in section 2.6 of @xcite .
notice that the successive heights @xmath293 of the vertices on the `` spine '' @xmath294 are distributed as a homogeneous poisson process on @xmath295 with intensity @xmath57 . keeping proposition [ prop : unif - meas - inv ] in mind
, we show by the next result why it is convenient to think of @xmath284 when studying a random geodesic ray in @xmath76 sampled according to @xmath296 .
[ lem : ctgwbias ] the pair @xmath297 follows the distribution @xmath276 . recall the bijection @xmath298 introduced in section [ sec : ctgwtree ] .
if we apply its inverse @xmath299 to the size - biased ctgw tree @xmath284 and then take the natural compactification , we obtain a random compact rooted tree @xmath300 called the size - biased reduced tree of parameter @xmath275 $ ] . a poissonian description of @xmath300 is given as follows .
first , under the mapping @xmath301 the geodesic ray @xmath291 in @xmath284 corresponds to a distinguished point at height 1 in @xmath300 that we will still denote by @xmath294 .
along the ancestral line of @xmath294 in @xmath300 , we keep the same notation @xmath302 for the branching points , with their respective heights @xmath303 increasing to 1 as @xmath304 .
it is immediate to see that @xmath305 is distributed on @xmath306 as a poisson process with intensity measure @xmath307 .
moreover , if we set @xmath308 , the sequence @xmath309 consists of i.i.d .
random variables with the density function @xmath310 over @xmath60 $ ] . for any @xmath82
, we write @xmath311 for the `` same '' random tree as @xmath59 with the distance @xmath64 multiplied by the factor @xmath84 , which means that the infinite discrete tree @xmath312 associated to @xmath59 remains the same while all the uniform random variables @xmath313 involved in the definition of @xmath59 are replaced by @xmath314 . conditionally on @xmath305 , independently to each branching point @xmath315 of height @xmath316 on the ancestral line of @xmath294 in @xmath300
, we graft a random number @xmath317 of independent trees , all distributed according to @xmath318 .
the random number @xmath317 is independent of those trees grafted at @xmath315 , and it has the same distribution as @xmath319 . as in the case of the size - biased ctgw tree @xmath284 , we assume the independence of @xmath317 and the trees at grafted @xmath315 among all different levels @xmath320
. schematic representation of the size - biased reduced tree @xmath300 when @xmath179,width=415 ] to finally get the size - biased reduced tree @xmath300 , we take the completion of the tree obtained above with respect to its natural distance ( see fig .
[ fig : size - bias - delta ] ) .
this compactification is equivalent to adding all the points on the boundary of the grafted trees .
all these points together with the distinguished one @xmath294 form the boundary @xmath321 , which is defined as the set of all points in @xmath300 that are at height 1 . the @xmath5-stable continuum random tree arises as the scaling limit of rescaled discrete galton watson trees with a critical offspring distribution in the domain of attraction of a stable distribution of index @xmath275 $ ] ( see e.g. the monograph @xcite of duquesne and le gall ) . under the same assumption ,
if we condition the critical galton watson tree to survive up to generation @xmath2 and rescale it by the factor @xmath48 , after letting @xmath26 , we get as limit an @xmath136-stable tree conditioned to reach height 1 ( see theorem 4.1 in @xcite ) , from which the reduced stable tree @xmath58 can be obtained as a pruned tree by keeping only the points at height less than 1 that possess a descendant that reaches height 1 .
the probabilistic structure of the size - biased reduced tree @xmath300 can also be derived from a spinal decomposition ( theorem 4.5 in @xcite ) of the conditional @xmath136-stable tree along the ancestral line of a typical point at height 1 . since this point of view is not needed for proving our main results , we omit the details . recall that the random variable @xmath322 is defined , from an electric network point of view , as the effective conductance between the root and the boundary @xmath66 in @xmath58 .
this is equivalent to defining @xmath322 as the mass assigned by the excursion measure away from the root for brownian motion ( with no drift ) on @xmath58 , to the set of trajectories that reach height 1 before coming back to the root .
this probabilistic definition of the conductance has the advantage of being easily generalized to a more general setting . if @xmath228 is an infinite continuous tree , we define its conductance @xmath323 between the root and @xmath324 as the mass assigned by the excursion measure away from the root for brownian motion with drift @xmath211 on @xmath228 , to the set of trajectories that go to infinity without returning to the root . by the correspondence between the ctgw tree @xmath76 and the reduced stable tree @xmath58
, we can verify that @xmath325 has the same law as @xmath322 .
see ( * ? ? ?
* section 2.3 ) for details . in section 2.3 of @xcite
, it is shown that for @xmath31 $ ] the law of @xmath326 is characterized by the distributional identity in the class of all probability measures on @xmath327 $ ] .
now let us take into account the martingale limit @xmath100 . from and the branching property of @xmath76
, it follows that the joint distribution of @xmath328 satisfies the distributional equation @xmath329 in which @xmath330 are i.i.d .
copies of @xmath331 , the integer - valued random variable @xmath332 is distributed according to @xmath61 , and @xmath67 is uniformly distributed over @xmath60 $ ] .
all these random variables are supposed to be independent . under @xmath19 , we define for every @xmath333 $ ] a positive random variable @xmath334 distributed as @xmath323 under the probability measure @xmath335 .
if the size - biased reduced tree @xmath300 is viewed as an electric network of ideal resistors with unit resistance per unit length , we will see that its effective conductance between the root and @xmath321 has the same law as @xmath336 . for this reason ,
we sometimes call @xmath336 the size - biased conductance .
[ lem : conduc - bias ] the random variable @xmath337 satisfies the distributional identity @xmath338 where in the right - hand side the random variable @xmath339 has density function @xmath310 over @xmath60 $ ] , the integer - valued random variable @xmath340 has the size - biased distribution of @xmath61 , and @xmath341 are i.i.d .
copies of @xmath132 .
all these random variables @xmath342 and @xmath341 are independent . the law of @xmath337 is determined by @xmath343= \e[\mathcal{w}\a\,g(\mathcal{c}\a)]$ ] for every nonnegative measurable function @xmath344 on @xmath46 . using ( [ eq : rde - joint ] ) under the same notation , we have @xmath345 & = & \e\bigg[(1-u)^{\frac{1}{\al-1 } } ( \mathcal{w}\a_{1}+\cdots+\mathcal{w}\a_{n_\al } ) \,g\bigg(\bigg(u + \frac{1-u}{\mathcal{c}\a_1+\cdots+ \mathcal{c}\a_{n_\al}}\bigg)^{-1}\bigg)\bigg]\\ & = & \e \bigg[n_\al ( 1-u)^{\frac{1}{\al-1 } } \mathcal{w}\a_{1 } \,g\bigg(\bigg(u + \frac{1-u}{\mathcal{c}\a_1+\cdots+ \mathcal{c}\a_{n_\al}}\bigg)^{-1}\bigg)\bigg].\end{aligned}\ ] ] defining @xmath340 as the size - biased version of @xmath332 , we see that @xmath346 = \e [ n_\al ] \cdot \e \bigg [ ( 1-u)^{\frac{1}{\al-1 } } \mathcal{w}\a_{1 } \,g\bigg(\bigg(u + \frac{1-u}{\mathcal{c}\a_1+\cdots+ \mathcal{c}\a_{\widehat{n}_\al}}\bigg)^{-1}\bigg)\bigg].\ ] ] recall that @xmath347=\frac{\al}{\al-1}=\e[(1-u)^\frac{1}{\al-1}]^{-1}$ ] .
the statement of the lemma thus follows if we let @xmath348 be a random variable independent of @xmath349 and @xmath341 , satisfying @xmath350= \e\bigg[\frac{\alpha}{\alpha-1}(1-u)^{\frac{1}{\alpha-1}}f(u)\bigg]\ ] ] for every nonnegative measurable function @xmath351 .
the law @xmath352 of the conductance @xmath127 has been studied at length in proposition 2.1 of @xcite .
we now discuss some similar properties of @xmath337 . for every @xmath353 , @xmath354 , @xmath355 and @xmath356
, we define @xmath357 so that can be reformulated as @xmath358 where @xmath359 are as in .
let @xmath360 be the set of all probability measures on @xmath361 $ ] and let @xmath362 map a probability distribution @xmath363 to @xmath364 where @xmath365 is distributed according to @xmath363 and independent of @xmath366 .
[ prop : c*-law ] we fix the parameter @xmath31 $ ] . 1 . the distributional equation ( [ eq : c*-rde ] ) characterizes the law @xmath367 of @xmath337 in the sense that , @xmath367 is the unique fixed point of the mapping @xmath368 on @xmath369 , and for every @xmath370 , the @xmath371-th iterate @xmath372 converges to @xmath367 weakly as @xmath373 .
the law @xmath367 has a continuous density over @xmath374 .
3 . for any monotone
continuously differentiable function @xmath375 , we have @xmath376+\frac{\al}{\al-1 } \,\e\big[g(\widehat{\mathcal{c}}\a)\big]=\,\frac{\al}{\al-1 } \e\big[g(\widehat{\mathcal{c}}\a+\mathcal{c}\a_2+\cdots+\cc\a_{\widehat n_\al})\big],\ ] ] where @xmath337 , @xmath340 , and @xmath377 are as in .
we define , for all @xmath378 , the laplace transforms @xmath379 $ ] and @xmath380\colonequals \int_1^{\infty } e^{-\ell x/2}\,\widehat \gamma_\al(\mathrm{d}x).\ ] ] then @xmath381 solves the linear differential equation @xmath382 5 . if @xmath383 , only the first moment of @xmath384 is finite , whereas all moments of @xmath385 are finite .
the results for @xmath386 are stated without proof in @xcite .
the arguments given below is similar in spirit to that of proposition 2.1 in @xcite , but due to the fact that @xmath286 has an infinite mean for @xmath387 , sharper estimates are necessary in the present setting .
\(1 ) first of all , the stochastic partial order @xmath388 on @xmath360 is defined by saying that @xmath389 if and only if there exists a coupling @xmath390 of @xmath363 and @xmath391 such that a.s .
it is clear that for any @xmath393 $ ] , the mapping @xmath394 is increasing for the stochastic partial order .
we endow the set @xmath395 of all probability measures on @xmath361 $ ] that have a finite first moment with the @xmath193-wasserstein metric @xmath396 \colon ( x , y)\mbox { coupling of } ( \sigma,\sigma')\big\}.\ ] ] the metric space @xmath397 is polish and its topology is finer than the weak topology on @xmath398 .
let us show that @xmath394 maps @xmath395 into @xmath398 for any @xmath7 $ ] .
for any @xmath399 and @xmath400 , the weighted harmonic - geometric means inequality says that @xmath401 we fix some @xmath402 and apply the preceding inequality to obtain @xmath403 it suffices to show that the right - hand side of the last display has a finite first moment provided that @xmath365 has a finite mean .
notice that by , @xmath404= \e\bigg [ \frac{\al}{\al-1 } ( 1-u)^{\frac{1}{\al-1 } } \frac{1}{u^r ( 1-u)^{1-r}}\bigg]\ ] ] is finite . on the other hand , since a.s .
@xmath405 and @xmath406 for all @xmath407 , @xmath408 & \leq & \e\bigg [ \frac{x+\mathcal{c}\a_2+\cdots + \cc\a_{\widehat{n}_\al}}{\big(\widehat{n}_\al\big)^r}\bigg ] \nonumber \\ & = & \frac{\al-1}{\al } \sum_{k\geq 2 } \frac{k \theta_\al(k)}{k^r } \big(\e[x]+\e\big[\mathcal{c}\a_2\big]+\cdots + \e\big[\cc\a_k\big ] \big ) .
\label{eq : finite - r}\end{aligned}\ ] ] it is shown in ( * ? ? ?
* proposition 2.1 ) that @xmath132 has a finite mean .
meanwhile , we know by stirling s formula that @xmath409 when @xmath304 . as @xmath410 , it follows that @xmath411 .
the sum in is finite and therefore @xmath412 has a finite first moment .
next , we observe that @xmath413 is strictly contractant on @xmath397 . to see this ,
let @xmath390 be a coupling between @xmath414 under the probability measure @xmath40 , and assume that @xmath415 and @xmath390 are independent under @xmath40 .
then @xmath416 and @xmath417 provide a coupling of @xmath418 and @xmath419 .
using the fact that a.s .
@xmath420 , we have @xmath421 by definition , @xmath422 & = & \frac{\al-1}{\al}\sum_{k\geq 2 } k\theta_\al(k)\ , \e\bigg[\frac{1-v_\al}{(1+(k-1)v_\al)^2}\bigg ] \\ & = & \sum_{k \geq 2 } \theta_\al(k ) \ , \e\bigg[\frac{k(1-u)^{\frac{\al}{\al-1}}}{(1+(k-1)u)^2}\bigg ] \\ & \leq & \sum_{k \geq 2 } \theta_\al(k ) \ , \e\bigg[\frac{k(1-u)}{(1+(k-1)u)^2}\bigg]\end{aligned}\ ] ] where @xmath67 is uniformly distributed over @xmath60 $ ] .
notice that for for any integer @xmath423 , @xmath424=1+\frac{k-1-k\log k}{(k-1)^{2}}<1.\ ] ] thus , taking expectation and minimizing over the couplings between @xmath363 and @xmath391 , we get @xmath425\mathrm{d}_{1}(\sigma,\sigma')\\ & \leq & \bigg(1+\mathbb{e}\big[\frac{n_{\alpha}-1-n_{\alpha}\log n_{\alpha}}{(n_{\alpha}-1)^{2}}\big]\bigg)\mathrm{d}_{1}(\sigma,\sigma ' ) \;=:\ ; c_{\alpha}\ , \mathrm{d}_{1}(\sigma,\sigma')\end{aligned}\ ] ] with @xmath426 .
so for @xmath31 $ ] , the mapping @xmath427 is contractant on @xmath398 and by completeness it has a unique fixed point @xmath428 in @xmath398 .
furthermore , for every @xmath429 , we have @xmath430 for the metric @xmath431 , hence also weakly , as @xmath373 .
however , @xmath432 is also a fixed point of @xmath427 according to .
the equality @xmath433 will follow if we can verify that @xmath428 is the unique fixed point of @xmath413 in @xmath369 . to this end
, it will be enough to show that @xmath430 as @xmath434 , for every @xmath435 .
let us apply @xmath427 to the dirac measure @xmath436 at infinity to see that @xmath437 where @xmath438 is a copy of @xmath339 being independent of @xmath439 and @xmath341 under @xmath40 .
we claim that @xmath440 has a finite first moment .
in fact , applying yields @xmath441 & = \ , \e \bigg [ \frac{\al}{\al-1 } ( 1-u)^\frac{1}{\al-1 } \bigg(u+\frac{1-u}{\widetilde v_\al^{-1}+\cc\a_2+\cdots+\cc\a_{\widehat n_{\alpha}}}\bigg)^{-1}\bigg ] \\ & \leq\ , \e \bigg [ \frac{\al}{\al-1 } \bigg(u+\frac{1-u}{\widetilde v_\al^{-1}+\cc\a_2+\cdots+\cc\a_{\widehat n_{\alpha}}}\bigg)^{-1}\bigg ] \\ & = \ , \sum_{k\geq 2 } k\theta_\al(k)\e\left [ \bigg(u+\frac{1-u}{\widetilde v_\al^{-1}+\cc\a_2+\cdots+\cc\a_k } \bigg)^{-1}\right ] .
\end{aligned}\ ] ] we integrate with respect to @xmath67 to obtain @xmath442 & = \,\e\bigg[\frac{\log \big ( \widetilde v_\al^{-1}+\cc\a_2+\cdots+\cc\a_k\big)}{1-(\widetilde v_\al^{-1}+\cc\a_2+\cdots+\cc\a_k)^{-1 } } \bigg ] \\ & \leq\ , 2\ , \e\left[\log \big ( \widetilde v_\al^{-1}+\cc\a_2+\cdots+\cc\a_k\big)\right].\end{aligned}\ ] ] using the inequality @xmath443 for @xmath444 , we see that @xmath445 \leq \e\left[\widetilde v_\al^{-1/2}\right ] + \e \left[\log \big(\cc\a_2+\cdots+\cc\a_k \big)\right].\ ] ] the random variable @xmath446 has clearly a finite mean .
meanwhile , for any @xmath447 , we have @xmath448 for every @xmath449 , and thus @xmath450\leq \frac{1}{r}\e\left[\big(\cc\a_2+\cdots+\cc\a_k\big)^r\right]\leq \frac{1}{r}\e\left[\cc\a_2+\cdots+\cc\a_k\right]^r.\ ] ] if we take @xmath451 , by the same arguments following we deduce that @xmath452\leq \frac{1}{r } \sum_{k\geq 2 } k\theta_\al(k ) \e\left[\cc\a_2+\cdots+\cc\a_k\right]^r < \infty.\ ] ] therefore , we get @xmath453 . by monotonicity ,
@xmath454 for every @xmath455 , and from the preceding results we have @xmath456 for every @xmath435 .
this implies that @xmath433 is the unique fixed point of @xmath413 in @xmath369 .
\(2 ) for every @xmath457 we set @xmath458 , and @xmath459 for every integer @xmath423 .
it follows from ( [ eq : c*-rde ] ) that , for every @xmath460 , @xmath461 , \label{eq : repart}\end{aligned}\ ] ] where @xmath332 is distributed according to @xmath61 . since @xmath462 for every @xmath463 $ ] and @xmath423
, we obtain from the last display that @xmath464 } \widehat f_{\alpha}(t ) = 1-\big(\frac{t-1}{t}\big)^{\frac{\al}{\al-1 } } a_0\a,\qquad \forall t\in[1,2],\ ] ] where @xmath465 \in [ 1,2^{\frac{\al}{\al-1}}).\ ] ] from and , we observe that @xmath466 is continuous on @xmath374 . recall that @xmath132 has a continuous density ( see ( * ? ? ?
* proposition 2.1 ) ) .
it follows that all functions @xmath467 are continuous on @xmath374 . by dominated convergence the function @xmath468\ ] ]
is also continuous on @xmath374 . using and again
, we see that @xmath466 is continuously differentiable on @xmath374 .
consequently @xmath469 has a continuous density on @xmath374 .
\(3 ) to derive the identity , we first differentiate ( [ eq : repart ] ) with respect to @xmath470 to see that @xmath471\ ] ] holds for @xmath472 .
let @xmath473 be a monotone continuously differentiable function .
we multiply both sides of the last display by @xmath474 and integrate for @xmath470 from @xmath193 to @xmath475 to obtain @xmath476 + \frac{\alpha}{\alpha-1}\mathbb{e } \big [ g(\widehat{\mathcal{c}}^{(\alpha)})\big ] = \e\big[n_{\alpha } \,g(\widehat{\mathcal{c}}^{(\alpha)}+ \mathcal{c}\a_{2}+\cdots+\mathcal{c}\a_{n_{\alpha}})\big],\ ] ] which gives .
\(4 ) taking @xmath477 for @xmath478 in ( [ eq : g - c - c * ] ) , we obtain by using the generating function of @xmath130 given in ( [ eq : gene - fct ] ) .
\(5 ) the first moment of @xmath337 is finite because @xmath479 .
taking @xmath480 in ( [ eq : g - c - c * ] ) yields @xmath481=\e\big[\widehat \cc\a\big]+\frac{\al}{\al-1}\e\big[\widehat n_\al-1\big ] \e\big[\cc\a\big]\,.\ ] ] for every @xmath482 , as @xmath483=\infty$ ] the second moment of @xmath337 is infinite .
finally , by proposition 6 in @xcite , all moments of @xmath484 are finite . using @xmath485 + 2\,\e\big[g(\widehat{\mathcal{c}}^{(2)})\big]= 2\ , \e\big[g(\widehat{\mathcal{c}}^{(2)}+\cc^{(2)})\big]\ ] ] with @xmath486 ,
one can easily show by induction that all moments of @xmath487 are finite .
the following proof of theorem [ thm : dim - continu ] for all @xmath275 $ ] proceeds in much the same way as in the case @xmath386 .
we skip some details , for which we refer the reader to section 2.5 in @xcite .
first , we have seen in proposition [ prop : unif - meas - inv ] that the probability measure @xmath276 on @xmath270 is invariant under the shift @xmath269 .
taking into account that @xmath488 , @xmath259-a.s .
, we can verify , in a similar fashion as in ( * ? ? ?
* proposition 2.6 ) , that the shift @xmath269 acting on the probability space @xmath489 is ergodic .
let @xmath490 denote the height of the @xmath2-th branching point on the geodesic ray @xmath105 .
for every @xmath149 , it holds @xmath491 where @xmath492 stands for the @xmath493-th iterate of the shift @xmath269 .
it follows from the ergodic theorem that @xmath494-a.s . and thus @xmath495-a.s .
@xmath496 one calculates as in the proof of proposition [ prop : unif - meas - inv ] to get @xmath497 for a fixed geodesic ray @xmath498 , we write @xmath499 for the @xmath500-st branching point on @xmath105 , and we set @xmath501 . using the ergodic theorem again , we have @xmath502-a.s . ,
@xmath503 where the limit is identified with @xmath504 the calculation can be done analogously as in display ( 18 ) of @xcite .
we only point out that we need @xmath505 , which is true because @xmath506 ( cf .
theorem i.10.2 in @xcite ) . by considering the ratio @xmath507 and taking @xmath26
, we derive that @xmath259-a.s .
, @xmath509 from which the convergence ( [ eq : loc - dim - unif ] ) readily follows .
then we turn to the harmonic measure @xmath267 and set @xmath510 . using the flow property of @xmath267 ( see lemma 2.3 in @xcite ) and the ergodic theorem , we obtain the @xmath495-almost sure convergence @xmath511 we can evaluate the preceding limit as @xmath512 putting together with ( [ ergodic1 ] ) and ( [ ergodic11 ] ) , we see that @xmath259-a.s .
@xmath508-a.e . ,
@xmath513 which implies that the convergence ( [ eq : loc - dim - harm ] ) holds with the limit @xmath514,\ ] ] where in the right - hand side we keep the same notation as in . finally , it remains to check that @xmath122 defined by satisfies @xmath515 .
since @xmath516 , an upper bound for @xmath122 is @xmath517,\ ] ] which equals @xmath518 $ ] by independence and the fact that @xmath519=1 $ ] . for any fixed @xmath520 ,
there exists a positive constant @xmath521 such that @xmath522 for every @xmath523 . for similar reasons as in , we have @xmath524 \leq \frac{\al m}{\al-1}+ \big(\e\big[\cc\a\big]\big)^r \sum_{k\geq 2 } \theta_\al(k ) k^{1+r } < \infty.\ ] ] on the other hand , from and the display following we also know that @xmath525 by concavity of the logarithm , @xmath526 this inequality is strict if and only if for some @xmath527 , @xmath528 since the latter property holds with positive probability under @xmath259 , @xmath529 by ( [ ergodic2 ] ) , the right - hand side of the last display is equal to @xmath113
. therefore , we have @xmath530 . according to its definition , @xmath531.\ ] ] since we can identify the latter with @xmath532 - 1 $ ] by taking @xmath533 in , we get the identity @xmath534 - 1 = \e[\cw\a \cc\a]-1 $ ] .
let us show the monotonicity of @xmath122 to complete the proof of proposition [ prop : dim - formula ] .
indeed , we will prove that the law @xmath367 of the size - biased conductance decreases for @xmath275 $ ] in the sense of the usual stochastic ( partial ) order .
observe that the function @xmath535 defined in is * decreasing with respect to @xmath536 , * increasing with respect to @xmath537 , * increasing with respect to @xmath538 for every @xmath407 .
since @xmath539 for every @xmath540 $ ] , the random variable @xmath339 is increasing with respect to @xmath275 $ ] for the stochastic order .
meanwhile , we deduce from that @xmath541= r - r(1-r)^{\al-1 } , \quad \forall r\in ( 0,1].\ ] ] putting it into the identity @xmath542}{1-r}\ ] ] and recalling that a.s .
@xmath543 , we get @xmath544 which implies that for every integer @xmath545 , @xmath546 by taking the derivative with respect to @xmath136 , we see that the size - biased offspring number @xmath286 is decreasing with respect to @xmath275 $ ] for the stochastic order . recall that the conductance @xmath132 is also decreasing with respect to @xmath275 $ ] for the stochastic order ( see section 2.4 of @xcite ) .
now we take @xmath547 .
according to the previous discussion , for any fixed @xmath548 , @xmath549 which implies that @xmath550 for any probability distribution @xmath363 on @xmath327 $ ] . by proposition [ prop :
c*-law ] , for any @xmath107 $ ] , the law @xmath367 of @xmath336 has no atom and the iterates of @xmath551 converge weakly to @xmath367 . as a result , @xmath552 is stochastically dominated by @xmath553 .
therefore , we have shown that the law of the size - biased conductance @xmath336 is decreasing with respect to @xmath275 $ ] for the stochastic order .
in particular , the first moment @xmath554 $ ] decreases with respect to @xmath107 $ ] and so does @xmath122 .
we give here the proof of proposition [ prop : dimension - explosion ] .
first of all , let us collect some facts about the riemann zeta function @xmath557 .
it is well - known that @xmath558 has a simple pole at @xmath559 with residue 1 .
then for the derivative @xmath560 , we have @xmath561 the following inequality for gamma function ratios ( see for example @xcite ) is also needed .
for every integer @xmath423 and every @xmath562 , @xmath563 one can thus find two postive constants @xmath564 such that for any @xmath565 and any @xmath423 , @xmath566 starting from and the fact that @xmath567 , we have @xmath568\leq \frac{\al}{\al-1}\e\big [ \log \big(\cc\a_1 + \cc\a_2+\cdots+\cc\a_{\widehat{n}_\al}\big)\big],\ ] ] where in the last expectation we introduced another copy @xmath569 of @xmath132 independent of all other random variables .
let @xmath570 be a sequence of independent random variables uniformly distributed over @xmath60 $ ] .
since each @xmath571 is dominated by @xmath572 for the stochastic partial order ( because of ) , @xmath573\leq \e\big[\log\big(u_1^{-1}+u_2^{-1}+\cdots+u_k^{-1}\big)\big].\ ] ] in @xcite , we have already shown that @xmath574 \leq \big(2+\frac{k}{k-1}\big)\log k,\ ] ] from which it follows that @xmath575 thus , together with and , we get @xmath576 for a lower bound , let us write @xmath577 - \e\big[\log \widehat{\mathcal{c}}\a\big].\ ] ] since @xmath567 , @xmath578 \geq \e\big[\log \widehat{n}_\al\big]= \frac{\al-1}{\al}\sum_{k\geq 2 } k\,\theta_\alpha(k)\log k.\ ] ] still using and , we see that @xmath579 and hence @xmath580 \geq c_1>0.\ ] ] on the other hand , let @xmath581 be a random variable of density function @xmath310 .
due to , the size - biased conductance @xmath337 is dominated by @xmath582 for the stochastic order , and thus @xmath583\leq \e[\log v_\alpha^{-1}]$ ] . however , we define the analytic function @xmath584 that interpolates the harmonic numbers @xmath585 .
an integration by parts shows that @xmath586.\ ] ] since @xmath587 converges to the famous euler - mascheroni constant as @xmath95 goes to @xmath588 , we have @xmath589=\lim_{\alpha\downarrow 1 } \,(\alpha-1 ) \e\big[\log v_\alpha^{-1}\big]=0.\ ] ] combining this with and , we obtain @xmath590 the proof of proposition [ prop : dimension - explosion ] is therefore completed .
we briefly introduce the general notion of discrete rooted ordered trees , which extends the case of infinite trees without leaves that we have seen in section [ sec : treedelta ] .
a discrete rooted ordered tree @xmath591 is by definition a subset of @xmath154 such that the following holds : 1 .
@xmath592 ; 2 . if @xmath593 , then @xmath594 ; 3 . for every @xmath595
, there exists an integer @xmath596 such that , for every @xmath597 , @xmath598 if and only if @xmath599 . the quantity @xmath600 above is called the number of children of @xmath601 in @xmath591 . a vertex with no child is called a leaf . from now on ,
we will just say tree instead of discrete rooted ordered tree for short .
we write @xmath231 for the ( non - strict ) genealogical order on @xmath591 .
a tree @xmath591 is always viewed as a graph whose vertices are the elements of @xmath591 and whose edges are the pairs @xmath602 for all @xmath603 . since the lexicographical order on the vertices is not really used in our arguments , we will not pay much attention to this order structure . for an infinite tree @xmath591
, we say it has a single infinite line of descent if there exists a unique sequence of positive integers @xmath604 such that @xmath605 for all @xmath149 .
the height of a tree @xmath591 is written as @xmath606 .
the set of all vertices of @xmath591 at generation @xmath2 is denoted by @xmath607 . if @xmath608 , the tree of descendants of @xmath601 is @xmath609\colonequals \{w\in \mathcal{v}\colon uw\in \mathsf{t}\}$ ] . let @xmath591 be a tree of height larger than @xmath2 , and consider a simple random walk @xmath610 on @xmath591 starting from the root @xmath240 , which is defined under the probability measure @xmath611 .
we write @xmath612 for the first hitting time of generation @xmath2 by @xmath365 , and we define the discrete harmonic measure @xmath613 supported on @xmath614 as the law of @xmath615 under @xmath611 . for every integer @xmath20 ,
we let @xmath616 be a discrete galton watson tree with critical offspring distribution @xmath9 , conditioned on non - extinction at generation @xmath2 , viewed as a random subset of @xmath617 .
we suppose that the random trees @xmath618 are defined under the probability measure @xmath19 .
we let @xmath619 be the reduced tree associated with @xmath616 , consisting of all vertices of @xmath616 that have ( at least ) one descendant at generation @xmath2 .
note that @xmath620 for every @xmath621 .
when talking about a reduced tree , we always implicitly assume that the correct relabeling of the vertices has been done to make it a tree without changing the genealogical order .
set @xmath622 the non - extinction probability up to generation @xmath2 for a galton watson tree of offspring distribution @xmath9 .
if @xmath623 is the slowly varying function appearing in ( [ eq : stable - attraction ] ) , it has been established in ( * ? ? ?
* lemma 2 ) that @xmath624 by the asymptotic inversion property of slowly varying functions ( see e.g. ( * ? ? ?
* section 1.5.7 ) ) , it follows that @xmath625 for some other function @xmath626 slowly varying at @xmath475
. * size - biased galton watson tree .
* we denote by @xmath627 a size - biased galton
watson tree with offspring distribution @xmath9 .
it is defined similarly as the size - biased ctgw tree @xmath284 .
we refer to lyons , pementle and peres @xcite for the details .
the unique infinite line of descent @xmath628 in @xmath627 is called its spine . if @xmath629 denotes a random variable distributed according to the size - biased distribution of @xmath9 , that is , @xmath630 for every @xmath236 , then the offspring numbers of @xmath631 are i.i.d
. copies of @xmath629 .
let @xmath632^{(n)}$ ] be the finite tree obtained from @xmath627 by keeping only its first @xmath2 generations , i.e. , @xmath633^{(n ) } \colonequals \{u\in \widehat{\mathsf{t}}\colon |u|\leq n\}.\ ] ] as shown in @xcite , the law of the random tree @xmath632^{(n)}$ ] is that of @xmath21 biased by @xmath634 .
moreover , conditionally given the first @xmath2 levels of @xmath627 , the vertex @xmath635 on the spine is uniformly distributed on the @xmath2-th level of @xmath627 .
for every integer @xmath149 , let @xmath632^n$ ] be the finite tree obtained from @xmath627 by erasing the ( infinite ) tree of descendants of the vertex @xmath635 . by convention
, the vertex @xmath635 itself is kept in @xmath632^n$ ] .
we emphasize that in general @xmath632^n \neq [ \widehat{\mathsf{t}}]^{(n)}$ ] , since the height of @xmath632^n$ ] can be strictly larger than @xmath2 .
we let @xmath632^{*n}$ ] be the reduced tree associated with the tree @xmath632^n$ ] up to generation @xmath2 , which consists of all vertices of @xmath632^n$ ] that have ( at least ) one descendant at generation @xmath2 .
notice that @xmath632^{*n}$ ] is also the reduced tree associated with @xmath632^{(n)}$ ] up to generation @xmath2 .
take a tree @xmath591 of height larger than @xmath2 and consider the new tree @xmath636 obtained by adding to the graph @xmath591 an edge between its root @xmath240 and an extra vertex @xmath637 .
we define under the probability measure @xmath638 a simple random walk @xmath365 on @xmath636 starting from the root @xmath240 .
let @xmath639 be the first hitting time of @xmath637 by @xmath365 , and for every integer @xmath640 , let @xmath641 be the first hitting time of generation @xmath493 ( of the tree @xmath591 ) by @xmath365 .
we write @xmath642 this notation is justified by the fact that @xmath643 can be interpreted as the effective conductance between @xmath637 and the @xmath493-th generation of @xmath591 in the graph @xmath636 , see e.g.
* chapter 2 ) . here
are some basic facts concerning the conductance of the reduced galton watson tree @xmath47 .
[ prop : cv - conductance ] suppose that the critical offspring distribution @xmath9 is in the domain of attraction of a stable distribution of index @xmath275 $ ] .
then it holds that @xmath644{(\mathrm{d } ) } \cc\a.\ ] ] recall that @xmath132 is the conductance of the continuous reduced tree @xmath58 .
given the convergence result of rescaled discrete trees @xmath49 towards @xmath58 ( proposition 3.2 in @xcite ) , the previous proposition can be shown in the same way as proposition 11 in @xcite . the following moment estimate for the conductance @xmath645 is lemma 3.9 in @xcite .
[ lem : lr - gw - condct ] for every @xmath646 , there exists a constant @xmath647 depending on @xmath84 and the offspring distribution @xmath9 such that , for every integer @xmath149 , @xmath648\leq k.\ ] ] in the proof of theorem [ thm : dim - discrete ] , we will need a rear - view variant @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] of the size - biased galton watson tree @xmath627 , which can be tracked back to the inflated galton
watson tree introduced by peres and zeitouni in @xcite .
let us explain succinctly how this backward size - biased galton
watson tree @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] is constructed .
more details can be found in section 3.4 of @xcite .
first , the random tree @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] has a unique infinite ray of vertices @xmath650 , referred to as its spine .
we fix a genealogical order on the spine by declaring that , for every @xmath20 , the vertex @xmath651 is at generation @xmath652 of @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] . under this rule
, @xmath653 is a child of @xmath654 , which is a child of @xmath655 , etc . to describe the finite subtrees in @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] branching off every node of the spine , we take an i.i.d .
sequence @xmath656 of random variables following the size - biased distribution of @xmath9 .
independently to each vertex @xmath657 , we graft @xmath658 independent copies of an ordinary @xmath9-galton
watson descendant tree , so that @xmath651 has exactly @xmath659 children : one of them , @xmath660 , is on the spine while the rest of them are the roots of independent @xmath9-galton watson trees .
last but not least , we define the genealogical order on @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] by keeping the genealogical orders inherited from the grafted galton watson trees and combining them with the genealogical order on the spine .
for instance , @xmath655 is an ancestor of any vertex in one of the subtrees grafted at @xmath654 .
the notion of generation for every vertex in @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] can also be defined in a consistent manner : for any vertex @xmath147 not on the spine , there is a unique vertex @xmath661 on the spine such that @xmath147 belongs to a finite subtree grafted at @xmath661 , then we say that the generation of @xmath147 in @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] is equal to @xmath662 plus the initial generation of @xmath147 inside the corresponding grafted tree . see figure [ figure : backgwbias ] for an illustration .
[ figure : backgwbias],width=377 ] different from the inflated galton
watson tree described in peres and zeitouni @xcite , the vertex @xmath663 in @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] has no strict descendant .
for every @xmath149 , let @xmath664 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}]^{n}$ ] be the tree obtained from @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] by only keeping the finite tree above @xmath651 .
the vertex @xmath651 is taken as the root of @xmath664 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}]^{n}$ ] .
we observe that @xmath664 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}]^{n}$ ] has the same distribution as the random tree @xmath632^{n}$ ] defined at the end of section [ sec : tree - discrete ] .
moreover , the root @xmath651 of @xmath664 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}]^{n}$ ] corresponds to the root @xmath240 of @xmath632^{n}$ ] , and the vertex @xmath663 in @xmath664 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}]^{n}$ ] corresponds to the vertex @xmath635 in @xmath632^{n}$ ] .
to prove theorem [ thm : dim - discrete ] , we will follow the same idea developed in @xcite for the finite variance case : the problem can be reduced to calculating a certain hitting probability for simple random walk on the backward size - biased galton watson tree @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] ( see proposition [ prop : red - theorem ] below ) .
since we fix a critical offspring distribution @xmath9 satisfying , the parameter @xmath275 $ ] is always fixed in this section . for simplicity , we assume that all the random trees involved below are defined under the same probability measure @xmath19 . recall that @xmath665 - 1>\frac{1}{\al-1}$ ] is the constant that appears in .
our first reduction goes from the conditional galton
watson tree @xmath21 to the size - biased galton watson tree @xmath627 . using a result of slack (
* theorem 1 ) , one can easily adapt the arguments at the beginning of section 4 in @xcite to see that convergence holds if for every @xmath35 , @xmath666^{n}}(x_{\tau_n}=\mathbf{v}_n)\leq n^{-\lambda_\al+\delta } \big ) = 1.\ ] ] furthermore , owing to the correspondence between the truncated trees @xmath632^{n}$ ] and @xmath664 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}]^{n}$ ] , we can rewrite the previous display as @xmath667^n}(x_{\tau_n}=\mathbf{u}_0)\leq n^{-\lambda_\al+\delta } \big ) = 1.\ ] ] in order to show , we denote by @xmath668 the generations of the vertices on the spine of @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] where there is at least one grafted tree that has a descendant of generation 0 .
this sequence of negative integers @xmath669 is listed in the strict decreasing order , and we set by convention @xmath670 . for every @xmath671
, we also set @xmath672 .
for every @xmath149 , let @xmath673 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}})$ ] be the index such that @xmath674 .
we need the next result to study the asymptotic behavior of @xmath675 .
[ lem : size - bias - gene ] let @xmath629 be a random variable distributed according to the size - biased distribution of @xmath9 .
then there exists a function @xmath676 slowly varying at @xmath677 such that for every @xmath447 , @xmath678=1-\al ( 1-r)^{\al-1}\widetilde l(1-r).\ ] ] moreover , we have @xmath679 where @xmath623 is the slowly varying function appearing in .
we need a nice karamata s representation of the slowly varying function @xmath623 appearing in .
indeed , @xmath623 can be expressed as @xmath680 where @xmath681 is a constant and @xmath682 is a continuous function on @xmath306 such that @xmath683 as @xmath684 . to see this ,
we follow slack @xcite to set @xmath685 and @xmath686 . since @xmath687 and @xmath688
, we know that @xmath689 is strictly positive . from the latter expression , we also deduce that @xmath690 is analytic and has monotone derivative for @xmath691 .
it follows from a result of lamperti ( * ? ? ?
* theorem 2 ) that @xmath692 we define @xmath693 which is continuous for @xmath694 and satisfies that @xmath683 as @xmath684 . integrating @xmath695 we obtain @xmath696 where @xmath681 as @xmath688 . the claimed representation for @xmath623 readily follows . as an immediate consequence , @xmath697 for @xmath691 . by differentiating , we see that @xmath698 = \sum_{k\geq 1 } k\rho(k ) r^{k-1 } & = 1-\al(1-r)^{\al-1 } l(1-r)- ( 1-r)^{\al } l(1-r)\frac{\varepsilon(1-r)}{1-r } \\ & = 1-\al(1-r)^{\al-1 } \widetilde l(1-r)\,,\end{aligned}\ ] ] where we have put @xmath699 . when @xmath700 , we get as @xmath701 . the asymptotic equivalence of @xmath676 and @xmath623 implies that @xmath676 also varies slowly at @xmath677 .
recall that for every @xmath703 , there are @xmath704 independent galton
watson trees grafted at @xmath705 in @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] .
consider the event that at least one of those trees grafted at @xmath705 reaches generation 0 , and let @xmath706 be the corresponding indicator function .
then by definition , @xmath707=\e\big[(1-q_{j-1})^{\widehat n-1}\big].\ ] ] applying and lemma [ lem : size - bias - gene ] to the latter formula yields @xmath708 since @xmath709 , we deduce that @xmath710=\sum\limits_{j=1}^n \big(1-\p(\epsilon_j=0)\big)\ , \sim \ , \frac{\al}{\al-1}\log n\ , , \qquad \mbox{as } n\to \infty.\ ] ] since @xmath711 are independent , we also have @xmath712 , and the @xmath713-convergence of @xmath714 to @xmath57 follows immediately .
the a.s .
convergence is then obtained by standard monotonicity and borel cantelli arguments .
given @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] , for every @xmath715 we write @xmath716 for the ( quenched ) probability measure under which we consider a simple random walk @xmath610 on @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] starting from the vertex @xmath705 . under @xmath716 , we denote by @xmath717 the hitting time of generation 0 by the simple random walk @xmath365 , and for every @xmath718 , we write @xmath719 for the hitting time of vertex @xmath720 . the following extension of ( * ? ? ? * proposition 14 ) compares simple random walks on the truncated finite tree @xmath664 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}]^n$ ] and on the backward infinite tree @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] .
we skip its proof because one can easily adapt the original proof of proposition 14 in @xcite to the present setting .
[ prop : red - to - backward ] for every @xmath35 , there exists an integer @xmath721 such that for every @xmath722 , we have @xmath723^n}(x_{\tau_n}=\mathbf{u}_0)\geq n^{-\lambda_\al+\delta}\big ) \leq 2^{\frac{2\al}{\al-1}}\ , \p\big(p^{\check{\mathsf{t}}}_{m_{k_n}}(x_{s_0}=\mathbf{u}_0 , s_0<\pi_{m_{k_n+1}})\geq n^{-\lambda_\al+\delta}/2 \big),\ ] ] and @xmath723^n}(x_{\tau_n}=\mathbf{u}_0)\leq n^{-\lambda_\al-\delta}\big ) \leq 2^{\frac{2\al}{\al-1}}\ , \p\big(p^{\check{\mathsf{t}}}_{m_{k_n}}(x_{s_0}=\mathbf{u}_0 , s_0<\pi_{m_{k_n+1}})\leq n^{-\lambda_\al-\delta } \big).\ ] ] under the probability measure @xmath19 , we set therefore @xmath725 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$ } } } } } ) \colonequals p^{\check{\mathsf{t}}}_{m_k}(x_{s_0}=\mathbf{u}_0 , s_0<\pi_{m_{k+1}})\ ] ] for every @xmath671 . by the definition of @xmath726
, there exists at least one subtree grafted to @xmath727 that reaches generation 0 .
the root of this subtree is necessarily a child of @xmath727 distinct from @xmath728 .
if such a subtree is unique , we let @xmath729 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}})$ ] be the probability that a simple random walk starting from its root reaches generation 0 before hitting @xmath727 . if there is more than one such grafted trees , we take @xmath730 to be the sum of the corresponding probabilities . we justify this definition by the fact that @xmath730 can be interpreted as the effective conductance between @xmath727 and generation 0 in the graph that consists only of the vertex @xmath727 and all the subtrees grafted to it .
we also set , for every @xmath671 , @xmath731 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$ } } } } } ) \colonequals p^{\check{\mathsf{t}}}_{m_k-1}(s_0<\pi_{m_k}),\ ] ] which is the probability that a simple random walk starting from @xmath728 reaches generation 0 before hitting @xmath727 .
we write in addition @xmath732 . a recurrence relation between @xmath733 and @xmath734
is obtained in section 4.1 of @xcite , which states that @xmath735 thus we define , for every @xmath736 , @xmath737 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$ } } } } } ) \colonequals \log \big ( 1+\frac{c_j+\ell_{j+1}}{\ell_j}-\frac{\ell_j}{\ell_j+c_{j-1}+h_{j-1}}\big).\ ] ] proposition [ prop : red - theorem ] ( and thus theorem [ thm : dim - discrete ] ) can be easily deduced from this key lemma .
in fact , on account of lemma [ lem : kn - asymptotic ] and lemma [ lem : sum - l2-conv ] , for any @xmath35 , the event @xmath739 holds with @xmath19-probability tending to 1 as @xmath26 . since for sufficiently large @xmath2 , the last event is included in @xmath740 the required convergence ( [ eq : red - theorem ] ) follows immediately . therefore , we are left with the task of proving lemma [ lem : sum - l2-conv ] .
let us begin with several auxiliary results .
for every @xmath671 , we denote by @xmath741 the number of independent @xmath9-galton watson trees grafted at @xmath742 that reach generation 0 in @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] .
[ lem : nb - graft ] as @xmath371 tends to @xmath475 , the conditional generating function @xmath743 $ ] converges to @xmath744 . thus according to
, the conditional distribution of @xmath741 given @xmath745 converges in distribution to @xmath319 . for fixed @xmath746 , by independence of the @xmath747 subtrees grafted at @xmath742
, we have @xmath748 = \e\big[(rq_{k-1}+1-q_{k-1})^{\widehat n_k -1}\big]=\e\big[(rq_{k-1}+1-q_{k-1})^{\widehat n -1}\big].\ ] ] it follows from lemma [ lem : size - bias - gene ] that @xmath749=1-\al(q_{k-1}(1-r))^{\al-1}\widetilde{l}(q_{k-1}(1-r)),\ ] ] where @xmath676 is the slowly varying function appearing in .
thus we obtain @xmath750 & = & \frac{\e\big[r^{i_k}\big]-\p(i_k=0)}{\p(i_k\geq 1 ) } \\ & = & 1- \al(1-r)^{\al-1 } \frac{(q_{k-1})^{\al-1}\widetilde l(q_{k-1}(1-r))}{\p(i_k\geq 1)}.\end{aligned}\ ] ] recall that it has been observed in the proof of lemma [ lem : kn - asymptotic ] that @xmath751 using and , one readily verifies that @xmath752 the proof is therefore finished
. note that if @xmath9 has a finite variance , the previous lemma becomes trivial because @xmath19-a.s .
for all sufficiently large @xmath371 , there will be a unique subtree grafted at @xmath727 that reaches generation 0 .
first , given any fixed integer @xmath755 , there exists some constant @xmath756 such that @xmath757 as @xmath758 it is clear that @xmath759= \sum_{k\geq 1 } k^{1+r } \rho(k)<\infty$ ] , and then @xmath760\leq \frac{1}{c}\e\big[(i_k)^r\big]\leq \frac{1}{c}\e\big[(\widehat n)^r \big]<\infty.\ ] ] for notational convenience , we define a random variable @xmath761 under @xmath19 having the conditional distribution of @xmath741 given @xmath745 .
since @xmath762\leq \sum_{n\geq 1 } r n^{r-1 } \p(i_k^*\geq n),\ ] ] it suffices to show that for any @xmath753 , we can find a constant @xmath763 such that for all @xmath371 sufficiently large and for all @xmath548 , @xmath764 to this end , we rewrite as @xmath765 and recall that @xmath766 using potter s bounds for @xmath676 and the previous convergence , we know that for any @xmath767 , there exists @xmath768 only depending on @xmath769 such that for every @xmath770 , @xmath771 and @xmath772 it follows that for every integer @xmath770 , @xmath773 from which one can deduce using the classical arguments for karamata s tauberian theorem and the monotone density theorem . for more details ,
we refer the reader to ( * ? ? ?
* section 1.7 ) .
* @xmath775 and @xmath776 are two positive random variables with the same distribution given by @xmath777 * take @xmath286 and @xmath778 two integer - valued random variables following the size - biased distribution of @xmath61 .
let @xmath779 and @xmath780 be two sequences of random variables identically distributed according to @xmath352 .
then @xmath781 * @xmath337 is distributed according to @xmath367 .
then notice that conditionally on @xmath726 and on the number @xmath783 of subtrees grafted at @xmath727 that reach generation 0 , these subtrees are independent @xmath9-galton watson trees conditioned to have height greater than @xmath784 .
recall that @xmath730 is the sum of the conductances between @xmath727 and generation 0 in those subtrees . on account of proposition [ prop : cv - conductance ] and
lemma [ lem : nb - graft ] , we can obtain the convergence in distribution of @xmath785 to @xmath786 by calculating the laplace transform of @xmath785 .
a slight change of this argument by first conditioning on @xmath787 gives the joint convergence @xmath788 which holds jointly with ( [ eq : ml - conv ] ) , provided we keep all the limiting random variables independent .
finally , recall that with the notation of section [ sec : cv - dis - cond ] , @xmath789 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}]^{m_{k-1}-1})$ ] .
the convergence @xmath790 can be shown by the same approximation method used in the proof of lemma 17 in @xcite .
since @xmath791 only depends on @xmath787 and the finite tree strictly above the vertex @xmath792 in @xmath649 { \rotatebox[origin = c]{180}{\vbox to 0pt{\hbox{$\widehat{\hphantom{\mathsf{t}}}$}}}}}$ ] , the last convergence holds jointly with ( [ eq : ml - conv ] ) and ( [ eq : c - conv ] ) , provided @xmath337 is taken to be independent of all the random variables involved in the definition of @xmath793 .
we fix @xmath795 . for any @xmath797
, there exists a positive constant @xmath798 sufficiently large so that @xmath799 for every @xmath399 . using this inequality together with lemma [ lem : lr - gw - condct ]
, we can easily adapt the proof of ( * ? ? ?
* lemma 18 ) to show the existence of positive constants @xmath800 and @xmath801 , both independent of @xmath371 , such that for all @xmath423 , @xmath802 \leq c+\xi \,\e\big[(m_{k-1}h_{k-1})^r\big].\ ] ] the uniform boundedness of the sequence @xmath803)_{k\geq 1}$ ] thus follows . 1 .
@xmath806 converges to @xmath807 in @xmath88 and in particular @xmath808= \e[q_{\infty}]$ ] .
we have the equality @xmath809 = \frac{\al-1}{\al}\lal$ ] .
it holds that
@xmath810<\infty \quad \mbox { and } \quad \sup_{i , j\geq 1 } \e\big[(q_iq_j)^2\big]<\infty.\ ] ] moreover , for any integer @xmath811 , we have @xmath812<\infty\,.\ ] ] \(i ) as in the proof of lemma 19 in @xcite , we know that @xmath813 for any @xmath82 , there exists a constant @xmath814 such that @xmath815 for every @xmath523 .
it follows that @xmath816 recall that conditionally on @xmath726 and @xmath783 , the quantity @xmath730 is the sum of @xmath783 i.i.d .
conductances of reduced galton
watson trees having height @xmath726 . by lemma [ lem : lr - gw - condct ] ,
the first moments @xmath817)_{n\geq 1}$ ] are uniformly bounded by a finite constant @xmath818 . if @xmath819 , we apply jensen s inequality to see that @xmath820= \e\big[\e[(m_kc_k)^r\!\mid\ !
m_k , i_{m_k}]\big]\leq \e\big[\e[m_kc_k\!\mid\ !
m_k , i_{m_k}]^r\big]\leq k\,\e\big[(i_{m_k})^r\big].\ ] ] then lemma [ lemma : i_k - frac - moments ] implies that for any @xmath753 , @xmath821<\infty.\ ] ] as in the proof of assertion ( i ) in ( * ? ? ?
* lemma 19 ) , one can also show that for any @xmath822 , @xmath823<\infty.\ ] ] hence , @xmath824 is bounded in @xmath825 for some @xmath826 .
but owing to lemma [ lemma:5rv - cv - in - law ] , @xmath806 converges in distribution to @xmath807 .
therefore , the uniform integrability of @xmath824 gives the convergence in @xmath88 .
\(ii ) for the calculation of @xmath809 $ ] , we set @xmath827 they are independent with the same law of density @xmath310 on @xmath60 $ ] . notice that @xmath828 and hence @xmath829=\e[\log ( v'_\al ) ] + \e\bigg[\log\bigg ( \cc\a_2+\cdots+\cc\a_{\widehat n_\al}+\frac{1}{\mathcal{r}_\al } + \bigg(v'_\al+\frac{1-v'_\al}{\widehat{\mathcal{c}}\a+ \widetilde \cc\a_2+\cdots+\widetilde\cc\a_{\widehat n'_\al}}\bigg)^{-1}\bigg)\bigg].\ ] ] using ( [ eq : c*-rde ] ) to rewrite the second term in the right - hand side , we obtain @xmath829=\e[\log ( v'_\al ) ] + \e\bigg[\log\big ( \cc\a_2+\cdots+\cc\a_{\widehat n_\al}+\frac{1}{\mathcal{r}_\al } + \widehat{\mathcal{c}}\a \big)\bigg].\ ] ] however , the relation @xmath830 gives @xmath831=\e\big[\log\big ( 1-v_\al+v_\al\big(\cc\a_2+\cdots+\cc\a_{\widehat n_\al}+\widehat{\mathcal{c}}\a \big)\big)\big]-\e[\log(v_\al)].\ ] ] as @xmath832=\e[\log ( v'_\al ) ] > -\infty$ ]
, we deduce that @xmath829= \e\big[\log\big ( 1-v_\al+v_\al\big(\cc\a_2+\cdots+\cc\a_{\widehat n_\al}+\widehat{\mathcal{c}}\a \big)\big)\big].\ ] ] now we apply ( [ eq : c*-rde ] ) again to see that @xmath833 has the same distribution as @xmath834 accordingly , we conclude by that @xmath829= \e\bigg[\log\frac{\widehat{\mathcal{c}}\a + \cc\a_2+\cdots+\cc\a_{\widehat n_\al } } { \widehat{\mathcal{c}}\a}\bigg]= \frac{\al-1}{\al}\lal\,.\ ] ] \(iii ) the last assertion can be analogously shown as assertion ( iii ) in ( * ? ? ?
* lemma 19 ) , by adapting the arguments there in the same way as we have done for assertion ( i ) . concerning the supplementary result ,
it can be similarly treated since one can adjust the power exponent @xmath84 in the inequality @xmath815 to be arbitrarily small .
we therefore omit the details .
we are now ready to finish the proof of lemma [ lem : sum - l2-conv ] . assertions ( i ) and ( ii ) of lemma [ lemma : auxiliary - l1 ] give @xmath835= \lim_{k\to \infty}\frac{1}{k}\sum\limits_{j=2}^k \e[q_j ] = \e[q_{\infty}]=\frac{\al-1}{\al}\lal.\ ] ] with all the ingredients prepared above , we can prove the inequality @xmath836\leq \big(\e[q_{\infty}]\big)^2\ ] ] in a similar manner as lemma 20 in @xcite . in the corresponding arguments ,
notice that for controlling the error of approximation , although lemma 18 in @xcite is now replaced by its weaker analog lemma [ lemma : lr - bdd - hk ] , we are able to compensate by to apply the hlder inequality as required . finally , the desired @xmath713-convergence follows from @xmath837 \big)^2\bigg ] \!\leq \limsup_{k\to\infty } \e\bigg[\big(\frac{1}{k}\sum\limits_{j=2}^k q_j\big)^2\bigg]-2 \e[q_{\infty } ] \lim_{k\to\infty}\e\bigg[\frac{1}{k}\sum\limits_{j=2}^k q_j\bigg]+ \e[q_{\infty}]^2 \leq 0.\ ] ] height 2pt depth -1.6pt width 23pt , _ typical behavior of the harmonic measure in critical galton watson trees _ , to appear in annales de linstitut henri poincar , probabilits et statistiques , available at . | we study the typical behavior of the harmonic measure in large critical galton
watson trees whose offspring distribution is in the domain of attraction of a stable distribution with index @xmath0 $ ] .
let @xmath1 denote the hitting distribution of height @xmath2 by simple random walk on the critical galton
watson tree conditioned on non - extinction at generation @xmath2 .
we extend the results of @xcite to prove that , with high probability , the mass of the harmonic measure @xmath1 carried by a random vertex uniformly chosen from height @xmath2 is approximately equal to @xmath3 , where the constant @xmath4 depends only on the index @xmath5 . in the analogous continuous model ,
this constant @xmath6 turns out to be the typical local dimension of the continuous harmonic measure . using an explicit formula for @xmath6 , we are able to show that @xmath6 decreases with respect to @xmath7 $ ] , and it goes to infinity at the same speed as @xmath8 when @xmath5 approaches 1 . * keywords . *
size - biased galton watson tree , reduced tree , harmonic measure , uniform measure , simple random walk and brownian motion on trees .
* ams 2010 classification numbers .
* 60j80 , 60g50 , 60k37 . |
kinetically constrained spin models ( kcsm ) are interacting particle systems which have been introduced and very much studied in the physics literature to model liquid / glass transition and more generally glassy dynamics ( see @xcite and references therein ) .
a configuration is given by assigning to each vertex @xmath2 of a ( finite or infinite ) connected graph @xmath3 its occupation variable @xmath4 which corresponds to an empty or filled site , respectively .
the evolution is given by markovian stochastic dynamics of glauber type .
each site with rate one refreshes its occupation variable to a filled or to an empty state with probability @xmath5 or @xmath6 respectively provided that the current configuration satisfies an a priori specified local constraint .
here we focus on two of the most studied kcsm , the east @xcite and fa-1f models @xcite on hypercubic lattices ( @xmath7 ) : the constraint at @xmath2 requires for east model its right nearest neighbour to be empty , for fa-1f model at least one of its nearest neighbours to be empty .
note that in both cases ( and this is a general feature of kcsm ) the constraint which should be satisfied to allow creation / annihilation of a particle at @xmath2 does not involve @xmath8 , thus detailed balance w.r.t .
the bernoulli product measure at density @xmath5 is an invariant reversible measure for the process .
both models are ergodic on @xmath9 for any @xmath10 with a positive spectral gap which shrinks to zero as @xmath11 corresponding to the occurrence of diverging mixing times @xcite .
several numerical works and approximated analytical treatments have shown that relaxation for both models occurs in a more and more spatially heterogenous way as density is increased ( see section 1.5 of @xcite and references therein ) . for example when measuring the persistence field @xmath12 which equals to one if site @xmath2 has never changed its state up to time @xmath13 and zero otherwise a clear spatial segregation is observed among sites with 0/1 values of @xmath14 at time scales corresponding to the typical relaxation time of the persistence function which corresponds to the spatial average of the persistence field .
more quantitatively , if one measures the spatial correlation function of this persistence field , a dynamical correlation length corresponding to the extent of these heterogeneities can be extracted .
this length increases as the density is increased .
the occurrence of these dynamical heterogeneites has lead to the idea that the dynamics of kcm takes place on a first - order coexistence line between active and inactive dynamical phases @xcite . in order to exploit this idea in @xcite the fluctuation of the dynamical activity @xmath0 defined as the number of microscopic configuration changes on a volume of linear size @xmath15 in the time interval
@xmath1 $ ] has been investigated .
the mean activity scales as @xmath16 where @xmath17 depends on the density and on the choice of the constraints , as we will detail in section [ models ] .
thus one could expect that the probability of observing a deviation from the mean value scales as @xmath18 with @xmath19 for @xmath20 as it occurs for the models without constraints .
however , as it has been observed in @xcite , due to the presence of the constraint it is possible to realize at a low cost a trajectory with zero activity by starting from a completely filled configuration and imposing that a single site does not change its state ( see section [ heuristics ] for a detailed explanation of the mechanism behind this phenomenon ) .
analogously one can obtain a smaller activity than the mean one by blocking for a fraction of time a single site . as a consequence of this sub - extensive cost for lowering the activity @xmath21 for @xmath22 .
for the same reason , the moment generating function @xmath23 is non analytic at @xmath24 with a discontinuous first order derivative @xcite . in this paper
, we investigate the finite size scaling of the first order transition .
our main results are estimates of the cost of phase coexistence between the active and inactive dynamical phases . from these estimates , the relevant scaling asymptotic in can be determined . for east and fa1f in one dimension ,
we prove ( theorem [ teo : phasetrans ] ) that @xmath25 satisfies @xmath26 if @xmath27 and @xmath28 if @xmath29 .
this shows that a transition in occurs for a value @xmath30 with @xmath31 . as a consequence ,
the scaling of the large deviations differs for increasing or decreasing the activity ( see theorem [ fluctu1 ] ) .
we also analyze the measure on the space - time configurations @xmath32 which corresponds to the conditional measure with a fixed activity @xmath33 where @xmath34 is the parameter conjugated to @xmath35 in legendre transform and prove ( theorem [ teo : condmes ] ) that depending on the value of @xmath36 this measure has very different typical configurations which can be interpreted as active and inactive dynamical phases : for @xmath27 , @xmath37 concentrates on trajectories with the mean activity and for @xmath29 , it concentrates on trajectories with zero activity .
finally , we investigate the higher dimensional cases and show that the finite size scaling depends not only on the dimension but also on the boundary conditions ( theorem [ linearityd>1 ] ) .
this leads to a variety of scalings for the large deviations when the activity is reduced ( theorem [ fluctu2 ] ) .
the east and fa-1f models in one dimension are glauber type markov processes on the configuration space @xmath39 where @xmath40 .
both models depend on a parameter @xmath5 , with @xmath10 , which we will call the _
density_. here we will consider the models in finite volumes @xmath41 $ ] and we will be interested in the thermodynamic limit @xmath42 .
we call @xmath43 the configuration space correspondent to @xmath44 and denote by greek letters @xmath45 the elements of @xmath46 .
then for any site @xmath47 we let @xmath48 be the value of configuration @xmath49 at site @xmath50 and we say that @xmath50 is _ empty _
( _ filled _ ) if @xmath51 ( @xmath52 , respectively ) .
the markov process corresponding to both models can be informally described as follows .
each site @xmath47 waits an independent mean one exponential time and then , provided the current configuration satisfies a proper local constraint , we refresh the value of the configuration at @xmath50 by setting it to @xmath53 with probability @xmath5 and to zero with probability @xmath6 . instead if the constraint is not satisfied nothing occurs
. then the procedure starts again .
the specific choice of the constraint identifies the model : for east one requires that the right nearest neighbour of @xmath50 is empty ; for fa-1f model one requires that at least one among the right and left nearest neighbours of @xmath50 is empty . in formulas
the constraint at @xmath50 is satisfied for east and fa-1f in the configuration @xmath49 iff @xmath54 and @xmath55 , respectively .
note that in both models the constraint is local ( it depends on the configuration on a finite neighborhood of the to - be - updated site ) and does not depend on the value of the configuration on the to - be - updated site .
both models belongs to the larger class of kinetically constrained models ( kcm in short ) , which have been introduced and widely studied in physics literature ( see for reviews @xcite ) . in order for the above description to be complete ,
we need to specify what happens at sites near the boundary .
a standard choice in statistical mechanics is to defined the dynamics in finite volume by imposing a fixed boundary condition . here
the choice of this boundary condition is very delicate , indeed due to the presence of the constraints both models are very sensitive to the specific choice of these conditions even on large volumes .
for example for the east model it is easy to verify that if we fix a boundary condition equal to one at @xmath56 , we start the evolution from @xmath57 and we let @xmath2 be the position of the rightmost zero of @xmath49 , then at any subsequent time site @xmath2 stays empty and sites @xmath58 $ ] stay filled . in this case we say that the configuration is _ frozen _ on @xmath59 $ ] meaning that under the evolution , the configuration on these sites remains unchanged . on the other hand , for a boundary condition equal to zero at @xmath56
, it can be easily verified that there is no site on which the configuration is frozen no matter which is the choice of the initial configuration . from the above observation it follows that in the case of filled boundary condition the configuration space ( even on finite volume ) is not irreducible .
indeed there exists configurations @xmath60 such that it is not possible to devise a path of elementary moves with strictly positive rates which connects @xmath61 to @xmath49 .
instead if we take an empty boundary condition the configuration space is irreducible .
this can be easily verified by constructing a path which completely empties any configuration starting from the right boundary .
analogously for fa-1f model if one imposes filled boundary conditions both at @xmath62 and @xmath56 the configuration space is reducible . on the other hand
any of the choices which has at least one empty site in the couple @xmath63 is sufficient to guarantee irreducibility . here for both models we will only be interested on choices of the boundary conditions which guarantee irreducibility ( and therefore ergodicity as we consider finite systems )
. note also that for both models , no matter which choice we perform for the boundary condition , the dynamics satisfy detailed balance with respect to bernoulli product measure @xmath64 at density @xmath5 , namely @xmath65 with @xmath66 ( this is a direct consequence of the above observed fact that the constraint at @xmath50 does not depend on the value of @xmath67 ) .
therefore @xmath64 is an invariant measure for the process and in the irreducible case this is the unique invariant measure .
let us now give a formal definition of these processes via the action of their generator @xmath68 on local functions @xmath69 .
we introduce @xmath70 where @xmath71 stands for the configuration @xmath49 changed at @xmath50 , namely @xmath72 and we let @xmath73\ ] ] with @xmath74 the function that encodes the constraint at site @xmath50 , namely @xmath75 ( @xmath76 iff the constraint at @xmath50 is ( is not ) satisfied . thus @xmath74 is model dependent and for the east model with frozen empty boundary condition at the right boundary we set @xmath77;\,\,\,\,\,\,\,\,\,r_n=1\ ] ] for fa-1f with empty boundary condition at the right and left boundary we set @xmath78;\,\,\,\,\,\,\,\,\,r_1=1,\,\,\,r_n=1\ ] ] for fa-1f with empty boundary condition at the right boundary and occupied boundary condition at the left boundary we set @xmath79;\,\,\,\,\,\,\,\,\,r_1=(1-\eta_2),\,\,\,r_n=1\ ] ] and finally for fa-1f with empty boundary condition at the left boundary and occupied boundary condition at the right boundary we set @xmath80;\,\,\,\,\,\,\,\,\,r_1=1,\,\,\,r_n=(1-\eta_{n-1 } ) .
\label{cfas}\ ] ] as already mentioned our analysis will focus on the above choices which are the only ones which guarantee ergodicity . therefore in the following when we refer to the east model , to fa-1f with two empty boundaries and to fa-1f with one empty boundary we mean respectively the choice , and ( by symmetry reasons the choices and for fa-1f are equivalent
therefore we never consider the case ) . also , when we state results referred to fa-1f model without further specifying the boundary conditions it means that these results hold for both the choices and .
note that the generator can be equivalently rewritten as @xmath81 with @xmath82 the local mean at site @xmath50 .
the object of interest of our work is the total activity @xmath83 where @xmath84 is the random variable which corresponds to the number of configuration changes on site @xmath50 during the time interval @xmath1 $ ] .
it is easy to verify that @xmath85 is a martingale and therefore @xmath0 satisfies a law of large numbers with @xmath86 where @xmath17 , which will be referred to in the following as the mean instantaneous activity , is defined as @xmath87 note that the definition is well posed since by translation invariance @xmath88 for @xmath89 $ ] . for east , we get @xmath90 , for fa-1f instead @xmath91 . here
we will be interested in the study of the generating function which controls the fluctuations of @xmath0 for a given @xmath15 @xmath92 where here and in the following @xmath93 denotes the mean over the evolution of the process and over the initial configuration which is distributed with the equilibrium bernoulli measure @xmath64 at density @xmath5 ( where the density @xmath5 is fixed by the rates ) . with a slight abuse of notation for any event
@xmath94 we will also denote by @xmath95 the probability of @xmath94 , namely we set @xmath96 .
the main result of this paper is that in the scaling @xmath97 a phase transition occurs for this generating function .
more precisely if we define @xmath98 then the following holds : [ teo : phasetrans ] consider east or fa-1f model in @xmath38 at any @xmath10 .
there exists @xmath99 and a constant @xmath100 such that * for @xmath27 it holds @xmath101 ; * for @xmath29 it holds @xmath102 .
+ as a consequence of this theorem , estimates on the large deviations for a reduced activity can be obtained .
[ fluctu1 ] consider east or fa-1f model in @xmath38 at any @xmath10 .
for any @xmath103 it holds @xmath104 we conjecture that in the @xmath105 can be replaced by @xmath106 . in the regime @xmath27 and @xmath29 , this follows from the proof of theorem [ teo : phasetrans ] .
theorem [ teo : phasetrans ] ( i ) and ( ii ) will be proved in section [ linearity ] and [ largealpha ] respectively .
theorem [ fluctu1 ] will be proven in section [ new ] .
we also analyze the measure on the space - time configurations defined as @xmath107 and theorem [ teo : condmes ] states that depending on the value of @xmath36 this measure has very different typical configurations . for any configuration @xmath108 , we call @xmath109 the trajectory of the markov process generated by @xmath68 starting at time zero from @xmath49
. then the following holds [ teo : condmes ] consider east model and fa-1f model in @xmath38 with @xmath10 .
then there exists @xmath99 and a sequence @xmath110 with @xmath111 such that * for @xmath27 @xmath112 @xmath113 * if @xmath29 @xmath114 @xmath115 theorem [ teo : condmes](i ) and ( ii ) will be proven in section [ condmes1 ] and [ condmes2 ] respectively where stronger results ( lemma [ usolemma ] and [ usolemma2 ] ) concerning the concentration of the number of particles and the activity on mesoscopic boxes ( and not on the whole volume ) will also be established .
as we already mentioned in the introduction , the occurrence of a phase transition for the activity large deviations was first discovered in @xcite , where it was shown that @xmath116 has a critical value at @xmath117 ( see the left part of figure [ fig : courbes ] ) .
we recall below the mechanism of this phase transition in the case of the one - dimensional east model ( the case of fa-1f model is analogous ) .
when @xmath118 , the activity is increased and the large deviation functional is expected to be smooth .
negative values of @xmath119 lead to a decay of the activity and the constraint will play a crucial role . to have no activity
, a possible strategy is to start at time @xmath120 from a configuration totally filled ( @xmath121 for all @xmath50 ) and to remain in this configuration at any time .
this can be achieved by preventing @xmath122 from flipping to 0 , indeed the site @xmath15 is the only site allowed to flip due to the constraints and if it is maintained equal to @xmath53 then the rest of the configuration is blocked .
this leads to the lower bound @xmath123 where @xmath124 stands for the cost of the initial configuration and the last term is the probability that a poisson process of intensity @xmath125 has no jump up to time @xmath13 .
after rescaling , this shows that @xmath126 for @xmath127 and since @xmath128 and @xmath129 is increasing in @xmath130 it follows that @xmath131 for @xmath132 . on the other hand by convexity @xmath133 and
therefore at @xmath24 the first order derivative of @xmath129 has a jump . as noted in @xcite , it is remarkable that the phase transition occurs at @xmath134 which corresponds to the unperturbed dynamics . thus one may wonder if the singularity of the large deviation functional would lead to specific properties of the constrained systems . in this paper
, we investigate the finite size scaling of this first order phase transition through the function @xmath135 introduced in . this refined thermodynamic scaling corresponds to a blow up of the region @xmath134 .
the results of theorem [ teo : phasetrans ] are depicted in the right part of figure [ fig : courbes ] . in a range of @xmath119 of order @xmath136
, we have shown that the transition is shifted from 0 .
= 5 cm [ c][l]@xmath137 [ c][l]@xmath119 = 6 cm [ b][l]@xmath135 [ c][l]@xmath34 [ c][l]@xmath138 [ c][l]@xmath139 [ c][l][c][c]@xmath140 to understand this , we first suppose that no transition takes place at 0 and that @xmath129 could be expanded ( analytically ) .
if this was the case then we would expect that for large @xmath15 @xmath141 \ , = { { \ensuremath{\mathbb a } } } { \alpha}+ { \small o \left ( \frac{1}{n}\right ) } \ , , \end{aligned}\ ] ] where we used that @xmath142 and @xmath143 is equal to the mean activity @xmath144 ( in fact @xmath129 is not differentiable at 0 , but its right - derivative is equal to @xmath144 ) . part ( i ) of theorem [ teo : phasetrans ] shows that this behavior persists for negative values of @xmath34 provided @xmath145 . in this regime , we expect that the total activity is shifted from its mean value @xmath146 by an order @xmath147 which does not scale with @xmath15 . for such small shifts of the activity ,
the system remains very close to its typical state ( when @xmath15 is large ) .
in particular , theorem [ teo : condmes ] asserts that the mean density is very concentrated close to its equilibrium value @xmath148 .
a phase transition occurs for smaller values of @xmath34 and @xmath149 becomes equal to the constant @xmath150 .
the estimate leads to the lower bound @xmath151 and thus it is too crude to justify the claimed behavior . indeed for @xmath152 , an activity of order @xmath153 will not contribute to the scaling limit , thus it is more favorable to leave a small portion of the system active as depicted in figure [ fig : interface ] instead of forcing the whole configuration to remain totally filled .
= 4 cm [ t][l]@xmath153 by analogy with equilibrium , one can interpret @xmath154 as a surface tension between the inactive and the active region ( per unit of time ) @xmath155 where @xmath156 means that the rescaled activity is close to 0 in the thermodynamic limit .
contrary to the strategy in , the interface between the inactive and the active region is now allowed to fluctuate and the probabilistic cost is lowered .
the surface tension @xmath154 can be obtained from a variational problem which is specified for fa-1f with two empty boundaries in section [ largealpha ] and for east and fa-1f with one empty boundary in section [ largealpha2 ] . however
, our results do not provide a complete description of the system and it remains to prove that the typical configurations look like figure [ fig : interface ] .
nevertheless , theorem [ teo : condmes ] ensures that in the inactive regime almost all the sites are equal to 1 .
this confirms the conjectured picture .
our results ( theorems [ teo : phasetrans ] and [ teo : condmes ] ) do not provide the entire phase diagram for the generating function @xmath149 ( only for @xmath157 $ ] ) .
however , this is enough to deduce ( see theorem [ fluctu1 ] ) the correct order of the scaling for the large deviations of the activity below the mean value @xmath158 , \qquad -\sigma(1-u ) { \;\leqslant\;}\lim_{n\to\infty } \lim_{t\to\infty}\frac{1}{t}\log \left { \langle}\frac{{\ensuremath{\mathcal a}}(t)}{nt}\simeq u { { \ensuremath{\mathbb a } } } \right { \rangle } < { \alpha}_0 { { \ensuremath{\mathbb a } } } ( 1-u ) \ , .\ ] ] this scaling is anomalous compared to the extensive scaling in @xmath15 of the unconstrained models .
we conjecture that there is a unique critical value @xmath159 and that the two regimes remain valid up to @xmath159 as depicted in figure [ fig : courbes ] , namely @xmath160 for @xmath161 and @xmath162 for @xmath163 .
this would imply @xmath164 this conjecture is supported by numerical simulations and we refer to @xcite for an account on these numerical results .
if this conjecture is verified , then theorem [ fluctu1 ] can be improved and the large deviations for reducing the activity would be given by @xmath158 , \qquad \lim_{n\to\infty } \lim_{t\to\infty}\frac{1}{t}\log \left { \langle}\frac{{\ensuremath{\mathcal a}}(t)}{nt}\simeq u { { \ensuremath{\mathbb a } } } \right { \rangle}=-\sigma(1-u ) \ , .\ ] ] = 5 cm [ c][l]@xmath165 [ c][l]@xmath166 = 6 cm [ c][l]@xmath167 [ c][l]@xmath34 [ b][l]@xmath159 in @xcite , it was suggested that the large deviation approach could provide a natural way to define a dynamical free energy characterizing glassiness .
it is currently an open question to understand if ( and how ) the dynamical phase transition in @xmath137 can lead to quantitative predictions on the model at @xmath134 .
equilibrium statistical mechanics could serve as a guide to clarify this issue .
indeed a similar phenomenon to the one depicted above for east and fa-1f occurs in the finite size scaling of the ferromagnetic ising model . for an ising model in the phase transition regime ( @xmath168 ) ,
a first order phase transition occurs in the magnetic field @xmath166 and the magnetization @xmath165 is discontinuous at @xmath169 ( see figure [ fig : courbes ising ] ) . on a finite domain ,
say a square of size @xmath15 , with external boundary conditions @xmath170 , then the magnetization @xmath171 is continuous and approaches the graph of @xmath165 . a finite size scaling @xcite shows that up to rescaling @xmath172 , the magnetization @xmath173 converges to a step function with a jump at a critical value @xmath174 ( see figure [ fig : courbes ising ] ) .
the shift of the transition is reminiscent of the shift for the constrained models and it can be understood as follows . for @xmath175 $ ] ,
the magnetization is slightly lowered but remains close to the magnetization @xmath176 imposed by the @xmath170 boundary conditions , then for @xmath177 the negative magnetic field forces a droplet of the @xmath178 phase which fills the system .
the creation of this droplet has a cost proportional to a surface order @xmath179 ( where @xmath180 is the surface tension term ) , but leads to an energy gain @xmath181 .
thus , the critical value is obtained for @xmath182 in this analogy , the magnetization plays a role similar to the activity and @xmath183 are the conjugate parameters .
even so a first order phase transition occurs for the ising model at @xmath166 equal 0 , it is known that the @xmath170 pure phase , i.e. the gibbs measure obtained from the @xmath170 boundary conditions after the thermodynamic limit , is well behaved and that the cumulants of the magnetization can be obtained by taking the successive derivative of the pressure for @xmath184 . as @xmath119 has no physical meaning ( contrary to @xmath166 )
, it is not clear from the mere knowledge of the first order phase transition how to deduce precise informations on the constrained dynamics at @xmath185 . in the previous sections we have considered one dimensional east and fa-1f models .
both models can be extended to higher dimensions in a very natural way .
let us set some notation .
let @xmath187 , then @xmath188^d$ ] and @xmath189 and let @xmath190 with @xmath191 $ ] be the euclidean basis vectors .
the east and fa-1f models in dimension @xmath192 at density @xmath10 are glauber type markov processes with generator @xmath193 which acts on @xmath194 exactly as in with the sum over @xmath50 running now on @xmath195 , with @xmath196 defined again as in and @xmath74 defined for the east model as @xmath197;\qquad r_i(\eta)=1 { \mbox{~otherwise~}}\ ] ] and for fa-1f model with completely empty boundary conditions as @xmath198 \,\,\,\,\,\forall j\in [ 1,d];\qquad r_i(\eta)=1 { \mbox{~otherwise~}}.\ ] ] as in the one dimensional case , the above choice of the boundary condition is the only ergodic choice for the east constraints , while in the fa-1f case any choice with at least one empty boundary site is ergodic
. we will be interested here in all the choices of the boundary conditions which correspond to requiring a completely empty hyperplane of linear size @xmath15 and dimension @xmath199 with @xmath200 $ ] ( @xmath201 corresponds to the choice of a single empty boundary site ) .
in short we will say that we consider a boundary condition of dimension @xmath199 in this case ( note that the only ergodic choice for east corresponds to a particular boundary condition of dimension @xmath202 ) .
note that as in the one dimensional case both dynamics satisfy detailed balance with respect to bernoulli product measure @xmath64 at density @xmath5 , namely @xmath203 with @xmath66 .
let @xmath0 be defined as in where now the sum runs over all sites inside @xmath204 .
as for the one dimensional case it is immediate to verify that @xmath0 satisfies the law of large number @xmath205 with @xmath144 defined as in for @xmath206 with @xmath207 $ ] for all @xmath208 $ ] .
note that @xmath144 coincides with the one for the corresponding one dimensional model at the same density .
let @xmath209 be defined as in and let the rescaled generating function @xmath210 and the measure @xmath211 be defined as @xmath212 @xmath213 note that the above definitions if we set @xmath38 and @xmath201 are compatible with the definitions used in section [ models ] for the one dimensional case , namely @xmath214 and @xmath215 with @xmath216 and @xmath217 defined by equations and respectively .
we stress that the finite size scaling depends on the choice of the boundary conditions .
we expect this to be the choice which leads to a phase transition for the generating function , as in the one dimensional case .
this conjecture , as will be further clarified by the proof of theorem [ poor ] , is related to the fact that in order to have no activity for a @xmath192-dimensional model a possible strategy is to start at time zero from a completely filled configuration and prevent all the @xmath218 sites which are in contact with the boundary empty set from flipping .
the @xmath192-dimensional east model corresponds to @xmath219 independent one dimensional east models so that @xmath220 for any @xmath192 . thus the phase transition results of theorem [ teo : phasetrans ] hold for @xmath221 and theorem [ teo : condmes ] applies as well in this case . for fa-1f
it is not immediate to generalize the one dimensional results since the model can not be decoupled into independent one dimensional fa-1f models . in this case
we prove [ linearityd>1 ] consider fa-1f model in dimension @xmath186 with boundary condition of dimension @xmath199 with @xmath200 $ ] .
there exists @xmath222 and a sequence @xmath110 with @xmath223 such that for @xmath27 @xmath224 @xmath225 @xmath226 [ fluctu2 ] consider east or fa-1f model in @xmath186 at any @xmath10 .
for any @xmath227 $ ] it holds @xmath228 where @xmath229 was introduced in theorem [ linearityd>1 ] .
these results correspond to those of theorems [ teo : phasetrans](i ) and [ teo : condmes ] ( i ) and [ fluctu1 ] in the one dimensional case . for the large negative @xmath36 regime ,
the result for dimension larger than one is much less precise .
[ poor ] consider fa-1f model in dimension @xmath186 with boundary condition of dimension @xmath200 $ ] .
for any @xmath230 there exists @xmath231 such that @xmath232 theorem [ linearityd>1 ] and [ poor ] will be proven in section [ genesmall ] and [ secpoor ] respectively .
in section [ var ] we recall some basic tools from the donsker - varadhan large deviation theory and we establish a variational formula for @xmath233 .
we underline that this formula is valid in any dimension . finally , in section [ subsec : sg ] we recall a result on the spectral gap of the generator for kcm which will be used in some of our proofs . the donsker - varadhan theory for large deviations @xcite will be a basic tool to derive our results .
fix the dimension @xmath192 and let @xmath234 be the dirichlet form corresponding to the generator @xmath235 which is defined on any function @xmath236 as @xmath237 for future use we note that the dirichlet form can be rewritten by using the definition as @xmath238 with @xmath239 . for any smooth function @xmath240
, we define the time average of @xmath241 over the process as @xmath242 where @xmath243 is a trajectory of the markov process starting at time zero from @xmath49 .
the dynamics is reversible with respect to the measure @xmath64 , thus the donsker - varadhan theory asserts that for any @xmath244 @xmath245 where we recall that @xmath246 is the mean over the evolution of the process and over the initial configuration which is distributed with the equilibrium bernoulli measure @xmath64 and the supremum is over the positive functions @xmath247 which satisfy @xmath248 .
note that the r.h.s of corresponds to the largest eigenvalue of the modified operator @xmath249 .
furthermore , if one defines the empirical measure in @xmath250 by setting for any @xmath251 and any @xmath252 @xmath253 donsker - varahdan theory establishes that the large deviation functional of the empirical measure is the dirichlet form @xmath254 .
thus if we let @xmath129 be any function from @xmath255 , for any @xmath256\subset \mathbb r$ ] it holds @xmath257 \right { \rangle}= \lim_{t\to\infty } \frac{1}{t}\log { \langle}\pi_t(\psi)\in[a , b ] { \rangle}= -\inf_{g : \nu(g)=1,~~ g{\;\geqslant\;}0\atop \nu(g \psi ) \in [ a , b ] } { \ensuremath{\mathcal d}}_n ( { \sqrt g } ) \ , .
\label{dv2}\ ] ] for any @xmath258 , we consider the modified dynamics obtained by rescaling the time by @xmath259 .
the generator reads @xmath260 which is again reversible with respect to @xmath64 .
then by evaluating the radon - nykodim derivative @xmath261 where @xmath262 and @xmath263 denote respectively the probability of the trajectory up to time t for the process evolving under @xmath68 and @xmath264 , we obtain @xmath265 with @xmath266 where we recall that @xmath0 is the total activity up to time @xmath13 and @xmath196 has been defined in .
this implies in particular for any function @xmath267 @xmath268 where here and in the following @xmath269 is the expectation over the modified dynamics , i.e. the mean over the initial configuration distributed with @xmath64 and over the evolution of the process with generator @xmath270 .
note that together with definition and formula ( @xmath270 is reversible with respect to @xmath64 thus donsker - varadhan theory applies ) , leads to the following variational formula @xmath271 where the supremum is taken over the positive functions @xmath272 such that @xmath273 .
finally , another tool which we will use is the knowledge of a positive lower bound uniform on @xmath15 for the spectral gap of @xmath274 which is defined as @xmath275 where @xmath276 is the variance w.r.t .
the invariant bernoulli measure @xmath64 on @xmath277 and the minimization is over the functions @xmath69 which are not constant . the following holds [ teogap ] consider east or fa-1f model in any dimension and with any choice of the boundary condition which guarantees ergodicity . for any @xmath10 there
exists @xmath278 such that @xmath279 from this result and definition it follows immediately that @xmath280 let @xmath3 be a generic connected subset of @xmath281 and consider fa-1f with a single empty site at the boundary on @xmath3 .
we call @xmath282 the generator and @xmath283 the corresponding spectral gap defined as in with @xmath284 substituted by @xmath285 then @xmath286 the result for the east model has been derived in @xcite and subsequently proven in @xcite for a larger class of constraints including fa-1f in any dimension with completely empty boundary conditions .
actually , the latter result could be easily derived by the result in @xcite without using the technique of @xcite . instead
, the result for fa-1f with generic boundary condition and on a generic graph has been derived in @xcite .
throughout this section , we consider east and fa-1f models in one dimension with mean density @xmath10 .
let us start by defining the coarse grained activity .
let @xmath287 be such that @xmath288 is integer and partition @xmath44 into boxes @xmath289 with @xmath290 $ ] of size @xmath287 , namely @xmath291 $ ] .
we define the activity in the interior of @xmath292 as @xmath293 where @xmath294 are the sites such that the corresponding constraints depend only on the configuration inside @xmath292 , namely @xmath295 for east and @xmath296 for fa-1f .
we also define the coarse grained density as @xmath297 fix @xmath298 , we define the _ activity - density _ associated to a configuration @xmath49 as @xmath299 where the mean instantaneous activity @xmath144 has been defined in formula and in order for the above definition to be well posed we restrict @xmath300 to the values such that @xmath301 and @xmath302 . in the rest of the paper in any result that uses these labels
we imply that @xmath15 and @xmath287 are integers chosen in order that @xmath288 is also integer and that the above restriction on @xmath300 is satisfied .
the main result of this section is lemma [ prop : bad blocksnew ] which states that the probability of finding under the empirical measure @xmath303 a density of boxes with activity - density label equal to zero , namely with activity different from the mean activity and from zero and/or particle density different from @xmath5 and one is suppressed exponentially in @xmath304 and @xmath288 . in other words
locally we are equilibrated either in the completely filled state or in the mean state identified by @xmath64 . instead lemma [
lem : variance 0 ] guarantees that the probability of finding a density of boxes with activity - density label equal to one , namely completely filled boxes , is suppressed exponentially in @xmath304 ( but not wrt @xmath288 ) . before stating and proving these results
we give some inequalities which immediately follow from the above definition of the labels and which will be used in the subsequent sections .
recall the definition of @xmath305 , then @xmath306 @xmath307 recall the definition of the empirical measure @xmath303 , see equation , we define the following events : fix @xmath308 , @xmath309 integers , @xmath310 $ ] and @xmath298 . then for @xmath311 we define the event @xmath312 by requiring a density at least @xmath313 of @xmath314 activity - density labels . in formulas
@xmath312 is verified iff @xmath315 for any integer @xmath316 we also define @xmath317 which is verified iff @xmath318 [ defw ] we stress that , even if it is not explicited for simplicity of notation , the event @xmath312 and @xmath319 depend on the choice of @xmath320 .
[ prop : bad blocksnew ] there exists @xmath321 such that for any @xmath322 , any @xmath258 and any integer @xmath309 provided @xmath323 it holds @xmath324 [ lem : variance 0 ] for any @xmath230 , there is @xmath325 such that for any @xmath326 , for any @xmath258 , any @xmath327 and any @xmath298 it holds @xmath328 where @xmath278 has been defined in proposition [ teogap ] .
the blocking mechanism described in section [ heuristics ] implies that @xmath329 thus the scaling in can not be improved . as a consequence of lemma [ prop : bad blocksnew ] , we will see that [ bad ] there exists @xmath330 s.t . for any @xmath331 , any @xmath332 and any @xmath333 with @xmath334 specified in lemma [ prop : bad blocksnew ] and any @xmath327 it holds @xmath335 we start by proving a preliminary result . for @xmath336 $ ] ,
let @xmath337 be the event that there exists at least one empty site inside the box @xmath292 and @xmath338 be the event that is verified iff @xmath339 and define the function @xmath340 as @xmath341 [ prop : bad blocks ] set @xmath342 .
there exists a constant @xmath343 such that uniformly in @xmath15 for any @xmath258 it holds @xmath344 for any @xmath345 @xmath346 then using we get @xmath347 where the supremum is over the @xmath69 such that @xmath348 and @xmath349 .
notice that @xmath350\right ) .\ ] ] the first term in can be bounded from above by @xmath351 where we use the fact that @xmath352 does not depend on the variables inside @xmath292 , @xmath353 does not depend on the variables outside @xmath292 and the fact that @xmath354 . on the other hand for the second term in we have @xmath355\right ) & { \;\leqslant\;}\nu\left ( \mathds{1}_{\mathcal{z}_{i+1}}(\eta)\left[\sqrt f-\nu_{b_i}(\sqrt f)\right]^2\right ) ^{1/2}\nu\left ( \left[\sqrt f+\nu_{b_i}(\sqrt f)\right]^2\right ) ^{1/2}\nonumber\\ { \;\leqslant\;}2~\nu\left ( \mathds{1}_{\mathcal{z}_{i+1}}\mbox{var}_{b_i}(\sqrt{f})\right ) ^{1/2 } \ , , \end{aligned}\ ] ] where to obtain the first inequality we upper bound @xmath356 by one and we use cauchy - schwartz while for the second inequality we use the fact that the event @xmath357 depends only on the variables inside @xmath358 , thus it is independent on the variables in the block @xmath289
. then we notice that @xmath359 guarantees the existence of ( at least ) one zero inside @xmath358 and we let @xmath360 be the position of the first zero starting from the right border of this box . thus @xmath361 and by letting @xmath362 and @xmath363 be the subset of @xmath364 to the left ( respectively right ) of @xmath365 , namely @xmath366 $ ] and @xmath367 we have @xmath368 where @xmath369 ( @xmath370 ) is the configuration restricted to @xmath371 ( @xmath372 ) and we use the product form of @xmath64 and , in the last passage , the convexity of the variance and the fact that @xmath373 .
then by using the spectral gap inequality for the model on @xmath362 with a frozen zero at the right boundary together with the expression of the dirichlet form we get ( recall that @xmath278 is the lower bound on the infimum over @xmath15 of the spectral gap of @xmath44 at density @xmath5 ) @xmath374 where we denote by @xmath375 the constraints for the model on @xmath362 with empty boundary condition on the right boundary , namely @xmath376 if @xmath377 and otherwise @xmath378
if we are considering east or @xmath379 if we are considering fa-1f . then we note that for any @xmath49 such that @xmath380 and which equals @xmath370 on @xmath372 , it holds @xmath381 for any @xmath382 .
thus we can insert into and use this observation to get @xmath383 then , , and yield @xmath384 and for @xmath385 the same upper bound can be obtained along the same lines ( actually easily because the boundary condition guarantees a zero at the right border of @xmath386 ) .
thus for any function @xmath247 s.t . @xmath387 and @xmath388 it holds @xmath389 \\ & = \frac{n}{k}\left[\gamma m+\frac { 2 \gamma^2 } { s_{\rho}\exp(\lambda ) } -\frac{k}{n}\sum_{i=1}^{n / k}\left(\exp(\lambda/2)\sqrt { { \ensuremath{\mathcal d}}_{k , i}(\sqrt f)}-\frac{\sqrt 2\gamma } { \sqrt s_{\rho}\exp(\lambda/2)}\right)^2\right]{\;\leqslant\;}\frac{n}{k}\left[\gamma m+\frac { 2\gamma^2 } { s_{\rho}\exp(\lambda)}\right]\nonumber\end{aligned}\ ] ] with @xmath390 the contribution to the dirichlet form coming from the sites in the box @xmath292 , namely @xmath391 then by using and optimizing over @xmath392 we get @xmath393 this completes the proof .
we are now ready to prove the main results of this section . [
proof of lemma [ prop : bad blocksnew ] ] we recall the definition for the function @xmath241 , where @xmath353 is the event which is verified iff @xmath394 .
thus from the definition of the activity - density labels it follows immediately that the probability of @xmath353 goes to zero as the size of the box , @xmath287 , goes to infinity and it is bounded from above by @xmath395 .
thus provided @xmath333 it holds @xmath396 .
thanks to these facts we can apply lemma [ prop : bad blocks ] with the choice @xmath397 to obtain that @xmath398 where @xmath241 is defined in . we will now prove that the following inequality holds for any @xmath399 @xmath400 then collecting and the proof of is completed .
we are therefore left with the proof of which immediately follows from the following observation .
let @xmath401 be such that @xmath402 and @xmath403 for all @xmath404 $ ] .
then there exists @xmath405 $ ] such that @xmath406 . in order to prove this statement we consider separately the case ( a ) @xmath407 and ( b ) @xmath408 . in case ( a ) the result holds
since @xmath409 implies @xmath410 and @xmath411 implies @xmath359 , thus @xmath412 . in case ( b )
we distinguish subcases ( b1 ) @xmath413 for all @xmath404 $ ] and ( b2 ) there exists at least one site @xmath404 $ ] such that @xmath414 . if ( b1 ) holds then @xmath415 ( since @xmath416 implies @xmath417 and @xmath418 implies @xmath419 ) and the desired result is proven . in case ( b2 )
if we let @xmath316 be the smallest index in @xmath420 $ ] such that @xmath421 then @xmath422 ( indeed @xmath423 and therefore @xmath424 and @xmath421 implies @xmath425 ) and again the desired result is proven .
[ proof of lemma [ lem : variance 0 ] ] by donsker - varadhan large deviation principle and the spectral gap inequality we get @xmath426 where the infimum is over the positive functions @xmath247 such that @xmath427 we will now show that under the latter constraint @xmath428 then by choosing @xmath287 sufficiently large so that @xmath429 by collecting and the desired result follows ( note that @xmath430 implies @xmath431 since we only have to deal with the case @xmath432 ) . we are therefore left with proving that under conditions the inequality holds . for each box @xmath292 with @xmath290 $ ] we define the coarse grained variable @xmath433 by @xmath434 and let @xmath435 and @xmath436 be the bernoulli product measure with density @xmath14 on @xmath437 .
the marginal of @xmath438 on the coarse grained variables is given by @xmath439 with @xmath440 where the sum is over the @xmath49 s compatible with @xmath441 .
then @xmath442 where we used the fact that for each fixed @xmath443 it holds @xmath444 and the concavity of the square root .
then from we get @xmath445 thus in order to prove it is sufficient to show that @xmath446 when @xmath236 is defined as in and @xmath247 satisfies conditions which imply @xmath447 the latter inequality implies that there exists ( at least ) a box @xmath448 with @xmath449 $ ] such that @xmath450 .
let @xmath314 be the rightmost box which verifies this constraint and rewrite each configuration @xmath443 via the couple @xmath451 with @xmath452 .
thus @xmath453 where @xmath454 where now @xmath455 denotes the product measure with density @xmath14 on @xmath456 .
by jensen inequality @xmath457 thus yields @xmath458 where we used for the first inequality , the fact that @xmath459 thus @xmath460 and that @xmath461 for the second inequality and finally the bound @xmath462 for the last inequality . thus is proven and the proof of the lemma is concluded . before proving the last result of this section , lemma [ bad ] , we state separately a result which will be also useful in other proofs .
[ easy ] consider east and fa-1f model in @xmath38 .
there exists @xmath463 s.t . for any @xmath15 if @xmath464 @xmath465 if @xmath466 @xmath467 by jensen inequality it holds @xmath468 then from the definition of the process and recalling it follows immediately that @xmath469 where @xmath470 and @xmath471 for east ; @xmath472 , @xmath473 for fa-1f with two zeros at the boundary ; @xmath472 , @xmath474 and @xmath471 for fa-1f with one zero at the ( right ) boundary .
thus @xmath475 and immediately follows from these bounds and inequality . from and lemma [ easy ] for any @xmath476 @xmath477 if instead @xmath466 by using the fact that @xmath478 we get @xmath479 the result immediately follows from the above inequalities and applying lemma [ prop : bad blocksnew ] .
we start by extending to the higher dimensional case the notion of coarse grained activity .
we let again @xmath288 be integer and partition @xmath204 into @xmath481 boxes of linear size @xmath287 .
we let @xmath292 be the boxes ( with @xmath482 $ ] ) numbered in such a way that for any @xmath483 $ ] there is @xmath484 $ ] such that @xmath358 is obtained by shifting @xmath292 of @xmath485 .
then we define the activity - density labels as in the one dimensional case and the event @xmath486 as in definition [ defw ] with @xmath288 substituted by @xmath481 .
the following holds [ ahh ] consider fa-1f model in dimension @xmath186 with any boundary condition which guarantees ergodicity and @xmath10 .
the results in lemma [ prop : bad blocksnew ] hold also for fa-1f in dimension @xmath192 if we substitute @xmath288 with @xmath481 .
the proof follows the same lines as for lemma [ prop : bad blocksnew ] with some new ingredients that we detail here .
since there is at least one empty boundary condition , there exists at least one box which has an empty site on its boundary .
we let @xmath314 be the index of the smallest such box .
then we let @xmath487 and notice ( along the same lines as for the unidimensional case ) that @xmath488 for any @xmath49 .
thus provided we can establish for @xmath241 the validity of lemma [ prop : bad blocks ] with @xmath288 substituted by @xmath481 we can conclude along the same lines as for the one dimensional case .
the validity of this modified lemma [ prop : bad blocks ] also follows along the same lines as the one dimensional case with a different point that we detail here .
recall that in one dimension under the event @xmath489 which guarantees that there exists at least one empty site in @xmath358 we identified the rightmost such site , which we denoted by @xmath360 . then under the event that @xmath490 we somehow extended the variance up to @xmath491 $ ] and used the positivity of the spectral gap on @xmath362 ( which is a volume of the type @xmath492 ) with fixed empty boundary condition at @xmath365 .
now instead under the event @xmath489 we number the sites of @xmath358 as @xmath493 in a way that @xmath494 is nearest neighbour of @xmath495 and then let @xmath360 be the empty site with the biggest label . then under the event @xmath496 we let @xmath497 $ ] and use the positivity of the spectral gap on @xmath362 with fixed empty boundary condition at @xmath498 . note that now @xmath362 is not an hypercube and the boundary condition is a single empty boundary condition even if we are considering the dynamics on @xmath44 with , for example , completely empty boundary conditions .
nevertheless we can use the positivity of the spectral gap on a generic connected graph with empty boundary condition ( see proposition [ teogap ] ) to bound again the variance on @xmath362 with the dirichlet form and then proceed as in the one dimensional case .
[ nonhopiu ] consider fa-1f model in dimension @xmath186 with any boundary condition which guarantees ergodicity and @xmath10 .
there exists @xmath499 such that @xmath500 the proof uses lemma [ ahh ] and follows exactly the same lines as the proof for lemma [ bad ] .
consider fa-1f model in dimension @xmath186 with @xmath10 and boundary condition of dimension @xmath200 $ ] .
then @xmath501 with @xmath502 defined in proposition [ teogap ] .
we detail the proof in the case @xmath503 , @xmath504 with the specific choice that all the sites in the boundary set which have first coordinate equal to @xmath56 are empty .
the other cases can be proven analogously . as in the proof of lemma [ lem :
variance 0 ] we start by applying donsker - varadhan large deviation principle which leads to @xmath505 where @xmath506 is the set of positive functions @xmath247 s.t .
@xmath507 and we added the index @xmath508 to explicitate the fact that we refer here to the dirichlet form of the two dimensional model . by the monotonicity of the rates for any function @xmath247 , one has @xmath509 where @xmath510 is the contribution of the @xmath511 line to the dirichlet form and @xmath512 is instead the dirichlet form of the one dimensional fa-1f model on the @xmath511 line with empty boundary condition on the right border ( note that @xmath513 is a function of the configuration on all the sites that do not belong to the @xmath50-th line ) and @xmath514 denotes the variance w.r.t .
the bernoulli measure restricted to the @xmath50-th line with the other variables held fixed .
the first inequality follows from the fact that for any site belonging to the @xmath511 line if the constraint is satisfied for the one dimensional model so it is for the two dimensional model ( but the converse is not true ) , indeed for any @xmath2 it holds @xmath515 .
the second inequality follows by using the spectral gap inequality for the one dimensional model .
then we notice that the condition implies that @xmath516 where @xmath517 stands for the activity label on the @xmath314-th one - dimensional box ( of size @xmath287 ) of the line @xmath50 .
thus we get @xmath518 where we denote by @xmath519 the set of positive functions which satisfy the conditions @xmath248 and @xmath520 following the lines of lemma [ lem : variance 0 ] we get for any @xmath521 $ ] @xmath522 and inserting this result in and yields the desired result provided @xmath287 is chosen sufficiently large so that @xmath523 .
in this section we prove theorem [ teo : phasetrans ] ( i ) which establishes the linearity of the moment generating function @xmath149 for fa-1f and east model in the small @xmath36 regime .
then , we prove theorem [ teo : condmes ] ( i ) and the stronger result lemma [ usolemma ] on the conditional measure . in section [ genesmall ]
we generalize these results to higher dimensions .
the key ingredints which will be used in these ( quite technical ) proofs are the main results obtained in the previous section , namely lemma [ prop : bad blocksnew ] and lemma [ lem : variance 0 ] . in this section
we will prove the following proposition [ altra ] there exists @xmath524 s.t . for any @xmath525 and @xmath526
there is @xmath527 such that for all @xmath528 it holds @xmath529 from this proposition , we deduce the result follows from the definition using the lower bound of lemma [ easy ] and the upper bound of proposition [ altra ] . we are therefore left with proving proposition [ altra ] .
we distinguish two cases .
* case @xmath530 .
* choose @xmath531 and @xmath287 such that @xmath532 with @xmath334 defined in lemma [ prop : bad blocksnew ] , then @xmath533 where the sets @xmath534 and @xmath535 were introduced in definition [ defw ] .
from it follows that @xmath536 where @xmath537 @xmath538 and @xmath539 by using the fact that @xmath530 , we get from , and @xmath540 @xmath541 and @xmath542 recall lemma [ prop : bad blocksnew ] then since we have chosen @xmath543 there is @xmath463 such that @xmath544 then recalling inequality , definition [ defw ] for the event @xmath545 and our choice @xmath546 , we get from @xmath547 since the event @xmath548 is a subset of the event @xmath549 with @xmath550 , by using again we get from @xmath551 where we used the fact that the event @xmath552 is a subset of the event @xmath553 . then by using lemma [ lem : variance 0 ] , it holds for any @xmath554 that @xmath555 thus collecting and we get @xmath556 since @xmath557 and @xmath558 , one gets ( for @xmath559 small enough ) @xmath560 thus under this hypothesis by using and collecting , and if we now set @xmath561 and @xmath562 we get the desired result .
* case @xmath466 .
* recall definition [ defw ] and the definition for @xmath563 .
then it holds @xmath564 where we let @xmath565 for @xmath566 , then @xmath567 .
thus via definition since @xmath568 , using lemma [ prop : bad blocksnew ] we get @xmath569 then using definition and recalling inequality we get @xmath570 thus by using and collecting and if we now set @xmath571 and @xmath562 we get the desired result . the main result of this section is the following lemma [ usolemma ] which
, as we will explain , is a stronger version of theorem [ teo : condmes](i ) .
[ usolemma ] let @xmath572 and @xmath573 be two sequences such that @xmath574 note that there are subsequences which verify the above conditions , e.g. the choice @xmath575 and @xmath576 .
for each @xmath15 let the activity - density labels be defined with @xmath577 and @xmath578 .
then there exists @xmath222 such that if @xmath27 it holds @xmath579 fix @xmath580 sufficiently large in order that it holds @xmath581 with @xmath334 defined in lemma [ prop : bad blocksnew ] ( this is possible thanks to the hypothesis @xmath582 ) .
we distinguish two cases .
* case @xmath583 . *
since @xmath217 is a measure from inequality it holds @xmath584 recall equation which defines the conditional measure @xmath217 .
if we apply proposition [ easy ] to bound the denominator and to rewrite the numerator we get @xmath585 @xmath586 where @xmath563 and @xmath587 are the functions that have been defined respectively in and and here we set @xmath588 .
then we use and the assumption @xmath589 to conclude that @xmath590 then by using , we get for @xmath591 @xmath592 by construction @xmath593 . thus for @xmath594 and @xmath15 large enough @xmath595 note that the threshold @xmath139 obtained here is not as sharp as in proposition [ altra ] . the proof
is then completed via , and .
* case @xmath596 .
* recall definition [ defw ] , then it holds @xmath597 since @xmath217 is a measure from inequality it holds @xmath598 recall equation which defines the measure @xmath217 .
applying proposition [ easy ] to bound the denominator and to rewrite the numerator , there is @xmath463 such that @xmath599 where @xmath563 was defined in and here we set @xmath600 .
thus together with the hypothesis on @xmath573 and @xmath572 imply that @xmath601 on the other hand , by using inequality and again proposition [ easy ] to bound the denominator and to rewrite the numerator of the conditional measure , we get with @xmath602 @xmath603 thus @xmath604 the result is then proved thanks to , and .
it is immediate to verify that the event @xmath605 implies that @xmath606 .
therefore is proven by using lemma [ usolemma ] and taking @xmath607 .
a similar argument leads to result .
the proof follows by using the results of section [ secbad ] along exactly the same lines as the proof of theorem [ teo : phasetrans](i ) and [ teo : condmes](i ) .
in this section we prove theorem [ teo : phasetrans ] ( ii ) . we analyze in detail the case of fa-1f with two empty boundaries in section [ genesmall1 ] and then we sketch how the proof is extended to fa-1f with one empty boundary and east model in section [ largealpha2 ]
. our proof will provide a variational characterization of the constant @xmath608 which appears in the theorem and which , as explained in section [ heuristics ] , plays the role of an interface energy .
we underline that this variational problem , and therefore the value of @xmath608 , depends not only on the choice of the constraints ( the interface energy for east and fa-1f at the same density are different ) but also on the choice of the boundary conditions ( the interface energy for fa-1f with one and two empty boundaries are also different , see remark [ remdiff ] ) . finally , we prove theorem [ poor ] for fa-1f in higher dimensions .
[ largealpha ] fix integer @xmath609 and let @xmath610 be the set of probability densities on @xmath611 such that @xmath612 with probability one , namely @xmath613 let @xmath614 with @xmath615 the dirichlet form of fa-1f model with two empty boundaries and define the interface energy @xmath154 as @xmath616 the definition is well posed thanks to the following lemma [ monotonicity ] .
[ monotonicity ] @xmath617 is non - increasing in @xmath618 and in @xmath619 . for any @xmath620 it holds @xmath621 .
therefore the limit @xmath622 exists .
the positivity of @xmath617 immediately follows from its definition .
let @xmath623 be the function s.t .
@xmath624 . set @xmath625 then @xmath626 and @xmath627 note that we have used the fact that the occupation variable at position @xmath53 is unconstrained .
analogously if we set @xmath628 then @xmath629 and @xmath630 . thus @xmath631 follows .
we split the proof of theorem [ teo : phasetrans ] ( ii ) into upper and lower bounds which are stated in the two following lemmas let @xmath608 be defined as in .
then for fa-1f model with two empty boundaries and any @xmath464 it holds [ lowereasy]@xmath632 [ upperdifficult ] let @xmath608 be defined as in . then for fa-1f model with two empty boundaries and any @xmath633 it holds @xmath634
then [ theorem [ teo : phasetrans ] ( ii ) for fa-1f model with two empty boundaries ] the result follows immediately from lemma [ lowereasy ] and lemma [ upperdifficult ] .
we are now left with the proof of the two above lemmas .
we fix @xmath287 and take @xmath635 .
let @xmath636 be the event which is verified iff @xmath637 . then from we
get @xmath638 by using the fact that on the event @xmath636 , it holds @xmath639 for any @xmath640 $ ] we get @xmath641 which together with yields @xmath642 from the donsker - varadhan large deviation principle it holds @xmath643 where in order to obtain the last inequality we proceed as follows .
denote by @xmath236 the function that belongs to @xmath644 and s.t .
. set @xmath646 then it can be verified that @xmath647 , @xmath648 and @xmath649 . thus follows . from
and we therefore obtain @xmath650 the result follows by taking @xmath15 to infinity and then @xmath287 to infinity . before proving lemma [ upperdifficult ] , we state and
prove an auxiliary result .
fix integers @xmath609 and @xmath651 and consider @xmath652 the set of probability densities on @xmath653 such that @xmath654 with probability at least @xmath655 , namely @xmath656 we will now prove that provided @xmath657 is sufficiently small the interface energy is well approximated by the infimum of @xmath658 restricted to @xmath659 .
more precisely [ prop : interface energy approx ] for any @xmath660 @xmath661 fix @xmath662 .
the dirichlet form can be bounded from below by @xmath663 we define a new probability density @xmath236 on @xmath653 @xmath664 with @xmath665 .
note that @xmath666 , thus @xmath236 belongs to @xmath667 furthermore , since @xmath247 belongs to @xmath668 one has @xmath669 .
note that the dirichlet form of @xmath236 satisfies @xmath670 decompose @xmath49 as @xmath671 where @xmath672 and @xmath673 , then @xmath674 then by using cauchy - schwartz @xmath675 where to obtain the last inequality we used the fact that @xmath247 belongs to @xmath676 .
thus @xmath677 similarly it can be verified that @xmath678 thus by using , , and we get @xmath679 where for the last inequality we used that , as noted above , @xmath236 belongs to @xmath680 and @xmath681 .
recall definition [ defw ] and set @xmath682 and choose @xmath683 with @xmath334 defined in lemma [ prop : bad blocksnew ] and let @xmath684 .
then the following inequality holds @xmath685 which implies @xmath686 where @xmath687 @xmath688 @xmath689 as in , @xmath690 is bounded by @xmath691 on the other hand since @xmath464 , we have @xmath692 we notice that the event @xmath693 implies that there exists at least one box @xmath290 $ ] such that @xmath694 .
thus in this box , there are at least 4 consecutive sites occupied with high probability @xmath695 let @xmath696 be the event that is verified if holds .
we get @xmath697 donsker - varadhan large deviation principle implies @xmath698 where we have set @xmath699 and @xmath700 . by using lemma [ prop : interface energy approx ] , noticing that @xmath701 and recalling the monotonicity property stated by lemma [ monotonicity ] , we obtain @xmath702 thus @xmath703 where @xmath704 . by using definition and inequality ,
we get @xmath705 note that @xmath706 is a subset of the event @xmath707 .
thus , we can use an estimate similar to in order to bound from above @xmath708 @xmath709 where @xmath710 is a constant .
combining and , we obtain @xmath711 where we used that @xmath712 .
recall @xmath713 and @xmath704 .
thus for @xmath714 , we can choose @xmath559 small enough and @xmath287 large enough such that @xmath715 sending @xmath716 and then @xmath717 , we get the desired result by collecting , , and . here
we explain how to extend the results of the previous section to the case of east and fa-1f model with one empty boundary , thus completing the proof of theorem [ teo : phasetrans ] ( ii ) .
we start by giving the definition of the interface energy @xmath608 for these models .
fix integer @xmath718 and consider @xmath719 the set of probability densities on @xmath720 such that @xmath721 with probability one , namely @xmath722 let @xmath723 , then we define the interface energy @xmath154 as @xmath724 the definition is well posed thanks to the following lemma [ monotonicity ] .
[ monotonicity2 ] let @xmath725 be either the dirichlet form of fa-1f with one empty boundary or the dirichlet form of east .
then @xmath726 is non - increasing in @xmath618 and it holds @xmath727 . therefore the limit @xmath728 exists .
let @xmath729 be the function in @xmath730 s.t . @xmath731 .
then set @xmath732
. then @xmath733 and , as for inequality , one can verify that @xmath734 which implies @xmath735 ( since @xmath736 and @xmath737 ) .
we will now state a result which allows to approximate the interface energy in the spirit of lemma [ prop : interface energy approx ] .
let for any integer @xmath718 and @xmath738 @xmath739 then [ energy approxbis ] for any @xmath660 @xmath740 the proof is analogous to the proof of lemma [ prop : interface energy approx ] , therefore we sketch only the main points .
let @xmath741 and set @xmath742 with @xmath743 .
then @xmath236 belongs to @xmath730 and it holds @xmath744 where the first inequality is obtained following the same lines as in lemma [ prop : interface energy approx ] . by using the above definitions and results we are now ready to prove theorem [ teo : phasetrans ] ( ii ) .
[ theorem [ teo : phasetrans ] ( ii ) for east and for fa-1f model with one empty boundary ] it is enough to prove that if @xmath608 is defined as in then for fa-1f with one empty boundary and for east the same inequalities as in lemma [ lowereasy ] and lemma [ upperdifficult ] with @xmath745 . in order to prove the lower bound one introduces the event @xmath636 which is verified iff @xmath746 . along
the same lines used to obtain on can verify that @xmath747 the lower bound follows again by taking @xmath15 to infinity and then @xmath287 to infinity . in order to prove the upper bound
we follow exactly the same lines as in lemma [ upperdifficult ] , the only difference being that in inequality now the event @xmath696 is substituted by an event @xmath748 which verified iff @xmath749 then donsker - varhadan principle yields @xmath750 given @xmath247 on @xmath46 we define a new function function @xmath236 on @xmath751 as follows @xmath752 then one can verify that it holds @xmath753 and if @xmath247 satisfies @xmath248 and @xmath754 then @xmath236 satisfies @xmath755 and @xmath756 , therefore @xmath236 belongs to @xmath757 .
therefore from we get @xmath758 [ remdiff ] fix @xmath10 and let @xmath759 and @xmath760 be the interface energies defined in formulas and by using the dirichlet form of fa-1f with two empty boundaries and one empty boundary respectively
. then it can be easily verified that @xmath761 .
we start by establishing a result which is stronger than theorem [ teo : condmes ] ( ii ) . [ usolemma2 ] consider east or fa-1f model in one dimension .
let @xmath575 and @xmath576 and for each @xmath15 let the activity - density labels be defined with @xmath577 and @xmath762
. then if @xmath763 with @xmath764 defined in lemma [ upperdifficult ] it holds @xmath765 fix @xmath580 sufficiently large such that @xmath581 with @xmath334 defined in lemma [ prop : bad blocksnew ] . since @xmath766 is a measure from inequality it holds @xmath767 where we set @xmath684 .
we get from lemma [ lowereasy ] @xmath768 where @xmath690 has been defined in and we used inequality . choose @xmath34 such that @xmath769 . for @xmath770
defined in with @xmath771 , one has by using for any @xmath15 large enough @xmath772 thanks to the hypothesis on @xmath572 , @xmath773 , @xmath573 , it holds @xmath774 and @xmath775 thus by using , and the proof is concluded .
the event @xmath776 implies @xmath777 thus the result follows by taking @xmath778 with @xmath573 defined in lemma [ usolemma2 ] .
recall that we have extended definition [ defw ] to the higher dimensional case by substituting @xmath288 with @xmath481 .
we start from inequality @xmath779 which leads to @xmath780 lemma [ nonhopiu ] guarantees that @xmath781 then we notice that if at time zero the configuration is completely filled and the clocks on all the @xmath782 sites which are unconstrained do not ring up to time @xmath304 then @xmath783
. therefore @xmath784 and by inserting this bound in the denominator of the definition of the measure @xmath211 and using inequality ( extended to dimension @xmath192 ) , we get @xmath785 provided @xmath786 and the result is proven by inserting and into .
as we already recalled in the introduction , in absence of constraints ( i.e. for the model defined by and with @xmath787 ) , the probability of observing a large deviation from the mean value scales as @xmath788 with @xmath789 for @xmath790 . in this section we will prove theorems [ fluctu1 ] and [ fluctu2 ] which establish that a different scaling occurs for the large deviations of the activity below the mean value in the presence of constraints .
let us start by the upper bound .
let @xmath764 be defined as in theorem [ teo : phasetrans ] , then @xmath791 therefore by taking the @xmath792 on the right and left hand side and using theorem [ teo : phasetrans ] ( i ) , we obtain and the desired upper bound . for the lower bound , we consider fa-1f with two empty boundaries ( the proof in the other cases is analogous ) .
the contribution to @xmath793 can be decomposed into the contributions coming respectively from the configuration changes during the time interval @xmath794 $ ] and @xmath795 $ ] . with probability ( w.r.t . the mean over the initial distribution @xmath64 and the evolution of the process ) which goes to one
as @xmath13 goes to infinity the first contribution goes to @xmath796 and , thanks to the reversibility of @xmath64 , the distribution at time @xmath797 is still @xmath64 .
then we can impose that during the second time interval @xmath795 $ ] the contribution to @xmath793 is at most @xmath798 by requiring that @xmath799 for any time @xmath800 in @xmath801 $ ] .
thus @xmath802 where @xmath803 is the event which is verified iff starting from @xmath64 it holds @xmath804 . as we did for the event @xmath636 in we get here @xmath805 and the proof is completed . the lower bound is derived in the same way by freezing the configuration during the time interval @xmath807 $ ] .
the probability of realizing this event can be bounded from below by the probability that the @xmath782 unconstrained sites ( those which are in contact with the empty boundary ) remain frozen equal to 1 .
we wish to thank l.bertini , b.derrida , v.lecomte , f. van wijland and l.zambotti for very useful discussions .
we acknowledge the support of the french ministry of education through the anr blan07 - 2184264 .
c.t acknowledges the support of anr dynhet and of the erc advanced grant ptrelss 228032
. n.cancrini , f.martinelli , c.roberto , c.toninelli , _ facilitated spin models : recent and new results _ in methods of contemporary mathematical statistical physics , lecture notes in mathematics , p.307 - 339 r.kotecky ed . , springer ( 2009 ) ; j.p .
garrahan , r.l .
jack , v. lecomte , e. pitard , k. van duijvendijk , f. van wijland , _ first - order dynamical phase transition in models of glasses : an approach based on ensembles of histories _
, j. phys .
a * 42 * 075007 ( 2009 ) | we consider two cases of kinetically constrained models , namely east and fa-1f models .
the object of interest of our work is the activity @xmath0 defined as the total number of configuration changes in the interval @xmath1 $ ] for the dynamics on a finite domain .
it has been shown in @xcite that the large deviations of the activity exhibit a non - equilibirum phase transition in the thermodynamic limit and that reducing the activity is more likely than increasing it due to a blocking mechanism induced by the constraints . in this paper , we study the finite size effects around this first order phase transition and analyze the phase coexistence between the active and inactive dynamical phases in dimension 1 . in higher dimensions ,
we show that the finite size effects are also determined by the dimension and the choice of boundary conditions .
_ mathematics subject classification : 60k35 , 82c22 , 60f10 _ _ keywords : kinetically constrained models , non - equilibrium dynamics , large deviations , glassy dynamics , interacting particle systems . |
over the past decade , more than 200 new planets outside the solar system were discovered @xcite .
because the majority of these planets were detected through precise radial velocity measurements , most of the known extra - solar planets have a mass comparable to that of jupiter and a tight orbit around a star within a hundred parsec from the sun .
habitable conditions for life may exist on moons around these giant planets or in other planetary systems that better resemble the solar system . given the high detection rate of new planets , the question of whether intelligent life exists elsewhere in the galaxy is particularly timely .
the search for extra - terrestrial intelligence ( seti ) has a long history @xcite . in 1960 ,
a targeted search of two nearby sun - like stars used the 25-meter radio telescope of nrao @xcite .
this project ozma @xcite was followed by other initiatives using radio telescopes @xcite and some seti programs continue today ( http://www.seti.org/ ) .
nasa s program _ the high resolution microwave search _ ( hrms ) included both a targeted search at individual stars as well as an all sky survey , but was cancelled by congress in the early 1990s @xcite .
funding for seti initiatives today originates from non - profit organizations such as the planetary society @xcite and the seti institute @xcite .
the seti institute has revived the targeted search component of hrms through project phoenix .
the serendip program at uc berkeley , as well as the joint venture between the seti institute and uc berkeley involving the construction of the _ allen telescope array _ ( http://astron.berkeley.edu/ral/ata/ ) , perform seti work on radio telescopes that are also used for astronomical studies .
most of the radio emission from our civilization on earth originates in the frequency range of 50400mhz .
table i summarizes the main sources of radio emission , including telecommunication and military radars .
the bandwidth of the signals is narrow , with a fractional frequency width of @xmath6 for tv and fm stations and @xmath7 for the military radars . from a distance , the broadcasted power as
a function of frequency would appear as narrow spectral lines . for practical purposes , the traditional seti searches focused on high radio frequencies that do not match the frequency range over which our civilization is most luminous ( see table i ) .
coincidentally , a new generation of radio observatories is being constructed around the globe with the primary goal of probing the redshifted 21 cm ( 1420 mhz ) emission from neutral hydrogen before it was ionized by the first galaxies .
as these observatories focus on the redshift range of @xmath815 , they cover the frequency range of 80300 mhz .
this frequency range happens to coincide with the band over which our civilization is most luminous .
coincidentally , these sensitive new instruments could be suited for seti work in this frequency band .
the goal of this paper is to characterize and motivate the use of these new observatories for seti studies .
seti work requires frequency resolution which happens to coincide with the design for one of these forthcoming observatories , the mileura wide - field array ( mwa ) .
we therefore use this experiment as a benchmark example to illustrate our main conclusions .
the targeted cosmic fluctuations in the 21 cm brightness temperature are small , around 1020 mk , depending on redshift and angular scale .
detection of these fluctuations is particularly challenging since the sky brightness temperature at the same wavelength is four orders of magnitude higher @xcite .
the faintness of the signal will drive the design of instruments to have large collecting areas . during the cosmic epoch of reionization ,
when the universe is partially ionized , the 21 cm signal has a swiss cheese " topology created by the bubbles of ionized hydrogen ( which are invisible at 21 cm ) surrounded by regions in which hydrogen is neutral and can thus be observed @xcite .
the characteristic bubble size grows during reionization and eventually reaches the large scale of tens of comoving mpc at the end of reionization , depending on the detailed characteristics of the sources of radiation and the distribution of the gas @xcite .
getting a good measure of the size distribution of these bubbles and other statistical properties of the fluctuations will require mapping tens to hundreds of square degrees ( necessary in order to probe many large bubbles at the end of reionization ) with at least arcminute resolution ( in order to capture the many small bubbles present at the beginning of reionization ) .
the removal of the strong synchrotron foreground will be achieved by subtracting the sky maps at slightly different frequencies @xcite .
for the cosmological signal , this is equivalent to slicing the hydrogen distribution at different redshifts and hence one should see a different map of its bubble structure , while the synchrotron foreground remains nearly the same .
given that 21 cm cosmology is a new well - funded frontier @xcite , a promising strategy for seti would be to piggyback on the radio observatories under construction . next we summarize the characteristics of mwa and other forthcoming experiments and consider their sensitivity for seti work .
several observational groups around the globe are constructing low - frequency arrays that will be capable of mapping the three - dimensional distribution of cosmic hydrogen in the infant universe .
these arrays are aiming to detect the long - wavelength ( redshifted 21 cm ) radio emission from hydrogen atoms , and map the three - dimensional distribution of cosmic hydrogen by making measurements over a wide range of frequencies .
the low - frequency demonstrator ( lfd ) of the mileura wide - field array ( mwa ; see http://web.haystack.mit.edu/arrays/mwa/ for details ) is currently under construction , with completion expected in early 2008 .
the mwa - lfd will cover the frequency range of 80300mhz in 8khz bins and include 500 dipole - based antenna tiles .
each 4m@xmath94 m tile will contain 16 dipole antennas .
altogether the total collecting area will be 8000 m@xmath10 at 150 mhz .
the tiles will be scattered across a 1.5 km region , providing an angular resolution of a few arcminutes .
mwa will be located in a radio - quiet site in western australia , called mileura .
other experiments whose goal is to detect 21 cm fluctuations from the epoch of reionization at @xmath11 include the low - frequency array ( _ lofar _ ; _ http://www.lofar.org_ ) , the primeval structure telescope ( _ past _ ; _ http://arxiv.org/abs/astro-ph/0502029_ ) , and in the more distant future expansions of mwa - lfd with the ultimate observatory being the square kilometer array ( _ ska _ ; _ http://www.skatelescope.org_ ) .
it is difficult to gauge the likelihood of success for the seti endeavor @xcite . as a result ,
ambitious and thus expensive projects whose sole purpose is seti are unlikely to be funded .
it is thus particularly fruitful to understand the synergies and complementarities of other scientific projects with various types of seti searches . to motivate the potential of this approach
we first summarize the expected sensitivity of mwa - lfd and other experiments . to assess what such detection levels could mean for seti we use the current radio emission by our civilization as an example and also compare with other past and future seti studies .
it is important to note that high - redshift 21 cm surveys will operate at different frequencies than past and current seti searches , making them complementary .
they will also have very different observing strategies and so they will explore a different part of the signal parameter space . forecasting
how the signals from another civilization might look like is not an easy proposition .
it is clear from the start that although one might be able to come up with educated guesses it is very possible that if we ever detect such a signal it will look different from our initial guesses .
for example , one might argue that the characteristics of the signal might be very different if the signal we receive was broadcasted intentionally by another civilization to announce its presence , a beacon " of sorts , or just the leakage of signals that the extra - terrestrials produce for their own communication either within their own planet or between spacecrafts .
whether we look for a beacon or unintentional leakage , the signal we are after must posses some characteristic that will enable us to distinguish it from naturally produced signals .
examples of such characteristics might be peculiar time variability and very narrow frequency structure centered on a frequency where there are no atomic or molecular lines .
a good starting point when considering leakage "- type signals is the signal from our own civilization .
a summary of the strongest emission component from the earth is given in table [ earth ] ( adopted from ref .
the strongest sources of emission in terms of their power are military radars ( @xmath12_{\rm int}\sim 10^{-5}$ ] ) , followed by tv stations and fm radio stations ( @xmath12_{\rm int } \sim 10^{-9}$ ] ) . in terms of power per hz , the strongest signal originates from tv stations .
these two signals are emitted in a fairly isotropic way .
if our civilization was using a telescope to beam those signals as a beacon or perhaps to communicate with a distant spacecraft , then the effective emitted power could have been much larger .
of course in that case only a correspondingly smaller fraction of distant observers would detect the signal at any given time .
it is possible then that the signal we are after , would be much stronger than our own emission but with a small duty cycle .
it is important to keep in mind that various physical processes would broaden the bandwidth or modulate the frequency of the emitted radio signal .
in particular the motion of the planet around its parent star and the spin of the planet , both introduce shifts .
the shifts are especially important for long integration times with narrow bandwidths . for the case of the earth ,
the spin velocity @xmath13 would introduce a shift of order , @xmath14 over the course of a day .
the envelope of this variability will be modulated by the orbital velocity @xmath15 around the sun over the course of a year with , @xmath16 the significance of these frequency drifts will depend on the duration of the observations which could be very long , of order months , in the context of 21 cm experiments .
if the transmitters are not distributed uniformly on the surface of the planet , the planet s spin will also introduce a modulation in the intensity of the signal .
of course if the signal originates from a beacon then it might be significantly beamed or pulsed .
the modulations of both intensity and frequency could eventually become clues that the emission originates from another civilization or at least from a planet around a star . in summary ,
the generic property of the expected signals are that they are narrow in frequency and that they might have interesting time variability as a result of both the orbital motion of the planet and its rotation around its axis , in addition to the variability introduced at the transmission .
other observing techniques should be used to follow - up on any radio detection of planets in an attempt to check whether it could host life as we know it .
the _ habitable zone _ is the range of orbital radii over which liquid water may exist on a rocky planet , so that life as we know it could develop there based on a similar network of chemical reactions to the one that existed on the early earth . obviously , the habitable zone straggles the distance of 1au for a sun - like star , but its location changes for other stellar types @xcite .
the first extensive discussion on habitability was provided by kasting et al .
existing observational techniques could find planets in the habitable zone in the future @xcite .
proposed space missions , such as the terrestrial planet finder ( http://tpf.jpl.nasa.gov/earthlike/earth-like.html ) and darwin ( http://sci.esa.int/science-e/www/area/index.cfm?fareaid=28 ) , focus on searches for signatures of life on extrasolar planets .
so far , no extrasolar planet was found to reside within the habitable zone of its parent star ( but see the indirect constraints in refs .
obviously , all planets which will be found to reside in the habitable zone should be searched with great sensitivity for radio signals , and vice versa .
the different search methods are complementary and one should use all probes at our disposal to constrain the physical conditions on such planets and determine whether they might support life .
this obviously includes spectroscopy of the parent star , which would yield its distance , spectral type , as well as the location of its habitable zone .
the 21 cm signal from high - redshift hydrogen is faint , leading to brightness temperature fluctuations of the order of @xmath17 mk on top of a sky brightness of @xmath18 k dominated by galactic synchrotron radiation @xcite .
the hii regions responsible for a large part of the signal subtend tens of arcminutes on the sky @xcite , and so large fields of view will be required to eliminate uncertainties due to cosmic variance .
the mwa - lfd and other future surveys will be designed to have large collecting areas to be able to detect the weak 21 cm signal , with fields of view that cover many square degrees on the sky . observing the 21 cm signal
will also require integrating on those fields for long periods of time .
finally the technique that will be used to distinguish the high - redshift 21 cm signal from various astrophysical foregrounds and radio frequency interference involves mapping the signal with high frequency resolution . as a result , _
future experiments will have extremely good sensitivity and could thus place strong limits on potential seti signals_. the extensive time coverage would allow the study of signals which might only be transmitted in our direction for a small fraction of the time .
the point source sensitivity of an interferometer composed of antennae of individual area @xmath19 and @xmath20 baselines that observes for a time @xmath21 with a bandwidth @xmath22 is , f_pss = c 2 k t_sys a_a , [ sens ] where @xmath23 is a constant of order unity that depends on the spatial distribution of baselines .
the system temperature , @xmath24 , is dominated by the synchrotron emission from our galaxy and is thus frequency dependent and variable across the sky . for the mwa - lfd ,
the point source sensitivity at @xmath25 mhz is for other frequencies can be found at http://web.haystack.mit.edu/arrays/mwa/lfd .
] , f_pss = 0.4 ( 8 khz t_o ) ^-1/2 , where the current design aims at @xmath26khz . for a given distance to the source @xmath27
, this formula can be turned into the minimum power required of the transmitter for mwa to see it , @xmath28 here it is assumed that the intrinsic bandwidth of the signal @xmath29 is much smaller than @xmath22 . in this regime
one gains by making @xmath22 smaller .
note that _ the power level in equation ( [ pm ] ) is only an order of magnitude larger than the present - day military radar emission from the earth ( table i ) even for a source which is 100 pc away_. it is plausible that the actual emission from other civilizations would be stronger .
it is also possible that the emission would be beamed or pulsed so that it would be visible to us for only a small fraction of the time .
the required minimum power scales as @xmath30 .
thus a duty cycle of @xmath31 requires a source to be merely 10 times closer for the same emitted peak power .
one way to increase the sensitivity is to reduce the bandwidth of the observations down to @xmath32 below which there is no additional gain .
figure [ distance ] shows the power required for a transmitter ( assuming isotropic emission ) to be detectable by mwa - lfd as a function of distance to the source .
the three solid lines show results for an hour , a day or a month of integration on a field . for a month of integration time
, the earth radio signal could be detected out to a distance of @xmath33 pc interior to which there are @xmath34 stars over the whole sky ( the field of view of mwa is about a percent of the sky ) .
however , only a fraction @xmath35 of the military radars may be beamed in our direction , where @xmath4 ( or 1.5 ) for a radar beam of solid angle @xmath5 that remains steady ( or sweeps ) across the sky .
we also plot the expected one - month sensitivity of lofar , mwa 5000 ( a future extension of mwa with ten times its collecting area ) and the ska which would be able to detect the earth out to a distance of 0.2 kpc .
we note that the use of the ska in the broader context of the exploration of life was discussed in ref .
@xcite .
another figure of merit that can be used to compare mwa with other previous and future seti searches is the flux , in @xmath36 , that must be arriving on earth for a source to be detectable , f_pss = 3 10 ^ -26 w m^-2 ( 8)^1/2 ( t_o ) ^-1/2 .
[ ht ] = 3.4 in the promise of future 21 cm surveys is evident from figure [ distance ] . as the collecting area increases
, the arrays will probe larger distances .
next we compare the sensitivity of mwa with other seti probes in the radio frequency range @xcite .
a very important difference between mwa and previous probes involves its sensitivity to lower frequencies .
most radio seti programs have searched above 1 ghz and around the unredshifted 21 cm line at 1.42 ghz .
there were several reasons to focus on this frequency range .
first , the sky brightness temperature is smaller above 1 ghz as the synchrotron emission from relativistic electrons in our galaxy becomes sub - dominant relative to the cosmic microwave background .
second , when looking for a beacon " type signal it could be argued that other civilizations might center their transmitters around the 1.42 ghz line with the hope of reaching astronomers using that line for astronomical observations .
it is difficult to gauge whether such an argument carries much weight .
of course , our own radio broadcasting is far greater at lower frequencies , so at least for the purpose of eavesdropping " on another civilization , lower frequencies might be more interesting .
the fact that our civilization makes much use of the lower frequency spectrum presents severe technical difficulties for seti programs trying to operate in this frequency range as they have to filter - out our own radio - frequency interference ( rfi ) .
thus , 21 cm cosmology is a case in which an unrelated science driver will open a new and potentially _ more suitable _
window for seti programs .
the interest in high - redshift 21 cm surveys means that there will be significant efforts to control rfi by , for example , placing the observatories in remote locations with the lowest rfi record ( such as china , australia , africa , or even the moon ) , as well as developing new filtering techniques for rfi and ionospheric noise .
obviously seti programs could benefit significantly from these technological developments .
a good summary of previous and planned seti programs can be found in ref .
@xcite . here
we will only comment on some of the most sensitive of those programs and focus our attention only on radio band examples . in doing so
, we would like to demonstrate that seti work with the mwa - lfd and other 21 cm surveys could have very competitive sensitivities . the seti programs carried out at arecibo and the parkes telescope in australia achieved the best sensitivities with their very large collecting areas .
for example , the phoenix project carried out by the seti institute reached a flux limit of @xmath37 by searching the nearest 600 stars at frequencies in the range 1.23 ghz with a frequency resolution of 1 hz . the serendip iv project out of uc berkeley also uses arecibo to monitor the sky visible from arecibo up to a flux of @xmath38 .
it covers the sky every three years and looks for recurring signals .
it explores the frequency range @xmath39 ghz with a 0.6 hz bandwidth .
a companion serendip search with similar characteristics operates from the parkes telescope .
the collecting area of arecibo is approximately @xmath40 m@xmath10 , roughly ten times that of mwa - lfd . by working at higher frequencies where the sky is less bright ,
the sensitivity of previous surveys was enhanced .
the mwa - lfd with its deep observations of certain fields will be competitive or even better than those previous surveys at least for steady signals .
production of detailed images from the epoch of reionization will require an increase in collecting area of future arrays by a factor of 10100 .
arrays of dipoles such as the mwa - lfd , have a much larger field of view , around 400 square degrees at 200 mhz , while a telescope like arecibo has a field of view of less that a tenth of a square degree .
the long observing times required for high - redshift 21 cm observations also open a new regime for seti observations .
the next generation of seti work will be carried out by the allen telescope array ( ata ) , a joint venture of the seti institute and uc berkeley .
the will include about three hundred and fifty 6 m dishes .
the plan is to survey of order @xmath41 stars three times in the frequency range of @xmath42 ghz over a period of 10 years .
this instrument has a collecting area similar to that of the mwa - lfd but a smaller field of view .
it is complementary to what could be done with proposed 21 cm arrays over a different range of frequencies .
again , the higher frequencies lead to a smaller sky temperature and thus to a better raw sensitivity in terms of @xmath43 .
but the larger collecting area of future extensions of the mwa - lfd will partially compensate for their larger system temperature .
the observing strategy of high - redshift 21 cm surveys , namely deep integration on selected fields , is also very different from the ata survey .
it will open a window to search for other types of signals , such as those that could be intense but with a short duty cycle , perhaps appearing for a short time once every rotation period of the planet .
it is very important to stress that all these different seti projects have fairly different characteristics in terms of frequency coverage , time spent observing any given source , or bandwidth of the observation . given our ignorance about the characteristics of the extra - terrestrial signal , it is certainly worth using all these different instruments for the search .
in this context , it is clear that the redshifted 21 cm observatories could attain unique sensitivities that were never realized before .
figure 1 indicates that integration times of months to years on upcoming low - frequency observatories will provide an unprecedented sensitivity to radio broadcasts from an earth - like civilization at galactic distances of 0.010.5 kpc , within a spherical volume containing @xmath44 stars , where @xmath4 ( or 1.5 ) for a radar beam of solid angle @xmath5 that remains steady ( or sweeps ) across the sky .
the challenge to detect redshifted 21 cm radiation from the infant universe will open a new opportunity for eavesdropping on the leakage of signals from galactic civilizations in the same band of radio frequencies that our civilization uses for communication .
such an opportunity never existed before .
seti work requires an optimized algorithm and software for extracting the artificial signal out of the data stream of upcoming observatories aiming to detect the cosmological 21 cm brightness fluctuations .
if not searched for , the desired seti signal may be lost .
processing of the data will be done by filtering - out unwanted foregrounds . in particular ,
the radio - communication signal from a faint earth - like source can be separated from the frequency structure of the redshifted 21 cm fluctuations through its time dependence .
the desired radio signal would appear as a series of spectral lines whose frequencies would be doppler shifting continuously on both the spin period of the brodcasting planet ( a day for the earth ) as well as its orbital period around the parent star ( a year for the earth ) .
the planet - star system resembles a spectroscopic binary in which one of the objects ( the planet ) is a test particle in the gravitational field of a companion object ( the star ) whose mass can be inferred from its emission spectrum through follow - up observations with optical - infrared telescopes . under these conditions , measurements of the planet
s doppler effect over a full orbit would determine the inclination , eccentricity and semi - major axis of the planet s motion around the star .
the inferred orbital parameters can then be used to estimate the temperature on the surface of the planet and to assess whether it is likely to support liquid water or life as we know it .
the optimization of the detection algorithm could also lead to changes in hardware or observing strategy with the goal of improving the prospects for the detection of a seti signal .
although mwa - lfd provided the benchmark example in this paper , our conclusions extend to other experiments such as lofar , past , or the ska . in parallel to developing the detection and analysis filter for these experiments , it would be prudent to compile an extended catalog of all nearby stars with known planets where low - frequency radio emission might be detectable .
this work will extend previous analysis @xcite to include additional stars where planets were found recently , as well as to span larger distances than those probed by other seti experiments .
we acknowledge support from the fqxi grant # rfp1 - 06 - 22 and harvard university funds for a.l . ;
m. z. is supported by the packard and sloan foundations , nsf ast-0506556 and nasa nng05gg84 g . we thank the anonymous referee for useful comments .
g. marcy , r. p. butler , d. fischer , s. vogt , j. t. wright , c. g. tinney , c. g. , & h. r. a. jones , progress of theoretical physics supplement , * 158 * , 24 ( 2005 ) ; see also http://www.ifa.hawaii.edu/uhnai/ppv.htm j. tarter , annu , rev .
, * 39 * , 511 ( 2001 ) .
w. t. sullivan iii , s. brown , c. wetherill , science , * 199 * , 377 ( 1978 ) .
b. lopez , j. schneider , & w. c. danchi , astrophys .
j. , * 627 * , 974 ( 2005 ) .
.characteristics of the largest components of radio emission from the earth ( adopted from @xcite and public web sites on _ bmews _ ) .
the power per solid angle @xmath45 along the beam of military radars ( _ bmews _ ) is larger by two orders of magnitude than the total power entry in the table . [
cols="^,^,^,^,^,^,^ " , ] | the question of whether intelligent life exists elsewhere is one of the fundamental unknowns about our universe . over the past decade
@xmath0 extra - solar planets have been discovered , providing new urgency for addressing this question in these or other planetary systems .
independently of this perspective , new radio observatories for cosmology are currently being constructed with the goal of detecting 21 cm emission from cosmic hydrogen in the redshift range @xmath1 . the radio frequency band covered by these experiments overlaps with the range of frequencies used for telecommunication on earth , a regime that was never explored with high sensitivity before .
for example , the low - frequency demonstrator ( lfd ) of the mileura wide - field array ( mwa ) , will cover in 8 khz bins the entire frequency range of 80300 mhz , which is perfectly matched to the band over which our civilization emits most of its radio power .
we show that this and other low - frequency observatories ( culminating with the square kilometer array [ ska ] ) will be able to detect radio broadcast leakage from an earth - like civilization out to a distance of @xmath2pc , within a spherical volume containing @xmath3 stars , where @xmath4 ( or 1.5 ) for a radar beam of solid angle @xmath5 that remains steady ( or sweeps ) across the sky .
such a radio signal will show - up as a series of narrow spectral lines that do not coincide with known atomic or molecular lines .
the high spectral resolution attainable with the upcoming observatories will allow to monitor the periodic doppler shift of the broadcasted lines over the planet s orbital period around the parent star .
determination of the parent star mass through observations of its spectrum could then be used to infer the inclination , semi - major axis and eccentricity of the planet s orbit .
this , in turn , will allow to estimate the temperature on the planet s surface and to assess whether it can support liquid water or life as we know it . |
the main goal of these sessions will be to understand the dynamics of a system exhibiting a continuum of periodic orbits when we add a small periodic forcing .
the most paradigmatic example is probably the perturbed pendulum ; however , such systems massively appear in real applications , specially in celestial mechanics .
this has given rise to classical problems exhibiting extremely rich dynamics , such as the restricted three body problem .
in these sessions we will see some theoretical results , but we will mainly visualize them through analytical and numerical computations for a particular example , the forced pendulum : @xmath0 where @xmath1 is a small parameter , @xmath2 and @xmath3 a periodic forcing : @xmath4 .
we will mainly see that , when @xmath5 the face portrait looks like figure [ fig : pendulum ] ( right ) and that , when @xmath6 looks like figure [ fig : pendulum ] ( left ) . we will learn theoretical and numerical techniques to compute the surviving resonant periodic orbits ( big holes in figure [ fig : pendulum ] left ) . +
although everything we will see will be generic for any @xmath3 , we will fix this forcing from now on say to @xmath7 ant @xmath8 becomes @xmath9 .
+ it will be useful to write the system as a first order one by increasing the dimension .
let @xmath10 , then we can write it as @xmath11
we will start with some exercises in order to understand the dynamics of the unperturbed pendulum ( @xmath5 ) .
consider the unperturbed equations of the pendulum by setting @xmath5 in eq .
: @xmath12 @xmath13 1 .
obtain the equilibrium points and state their type .
2 . obtain a hamiltonian for system . 1 .
the system possesses an elliptic equilibrium point at the origin and two saddle points at @xmath14 .
the function @xmath15 is the hamiltonian of system up to a constant value .
write a small program in matlab to plot the phase portrait of system .
see @xcite . in this session
we will focus on what happens with the periodic orbits inside the homoclinic loop when we add a small periodic forcing to system ( say @xmath16 ) . to this end
, it will be crucial to compute the periods of the periodic orbits for @xmath5 .
let s us now see a general theoretical approach to obtain a formula .
assume we have hamiltonian system of the form @xmath17 where @xmath18 is a potential energy .
assume that we have a periodic orbit at level of energy @xmath19 given by @xmath20 .
then , we can compute its period , @xmath21 , as follows : @xmath22 using that the system is hamiltonian and , hence , @xmath23 we get @xmath24 isolating @xmath25 from @xmath20 , @xmath26 but now we have two problems .
first is that solving such an integral explicitly can be a nightmare , if possible at all .
second , doing integrals numerically is a difficult task , it s slow and imprecise . fortunately , there is an alternative : compute a poincar map using a transversal section to the periodic orbit and capture the return time .
write a small program in matlab to compute the periods of the periodic orbits using the poincar map from the section @xmath27 to itself . in general
, for autonomous systems , in order to study the existence of periodic orbits one typically considers the poincar map : @xmath28 where @xmath29 is a co - dimension one section . in our case
, we could take the vertical axis @xmath30 .
then , fixed points of the poincar map , points @xmath31 such that @xmath32 , become initial conditions for periodic orbits for the flow , whose period is the flying time .
periodic points of the poincar map , @xmath33 , also give rise to periodic orbits for the flow , which cross @xmath34 times the poincar section and their period becomes the addition of all flying times between consecutive impacts with the poincar section before @xmath35 is reached again .
+ however , if the system is not autonomous ( but periodic ) , then one needs to take into account the initial time and consider poincar sections of the form @xmath36 , with @xmath37 .
then , a sufficient condition for the existence of a periodic orbits becomes @xmath38 .
that is , after the point @xmath35 is reached again after @xmath34 crossings and the total spent time is a multiple of @xmath8 .
then , the flow possesses an @xmath39-periodic orbit crossing the section @xmath34 times .
note that , if total time spent to reach @xmath35 again is not a multiple of @xmath8 , then nothing can be said about the existence of a periodic orbit .
one of inconveniences of using poincar maps with non - autonomous systems is the need of computing the flying time .
although this can be done numerically , it requires extra computations than simply numerically integrating a flow , as one needs to compute the crossing with the section .
alternatively , one can use the stroboscopic map , which consists of integrating the system for a time @xmath8 : @xmath40 where @xmath41 then , if @xmath42 , then @xmath35 is the initial condition for a @xmath39-periodic orbit : @xmath43 write a small program in matlab to compute the stroboscopic map . write also a script to iterate it several times ( say @xmath44 ) for different initial conditions at the @xmath25 axis . note the resonances
( probably you will need to play with @xmath45 to observe things better ) .
first use @xmath5 , and then increase it a little bit and see what happens .
let us consider a planar field of the form @xmath46 where @xmath47 @xmath1 is a small parameter and @xmath3 is @xmath8-periodic in @xmath48 : @xmath49 for simplicity , let us assume that , for @xmath5 , the unperturbed system @xmath50 is hamiltonian .
that is , there exists a function @xmath51 such that @xmath52 moreover , let us assume that the unperturbed system satisfies the following : 1 .
there exists a compact region completely covered by a continuum of periodic orbits .
2 . let @xmath21 be the period of the periodic orbit located at the energy level @xmath53 : @xmath54 assume that @xmath55 .
then we have the following result : [ theo : melnikov ] assume the above conditions are satisfied .
assume that the unperturbed system has a periodic orbit , @xmath20 , of period @xmath56 where @xmath8 is the period of the periodic forcing .
let @xmath57 be an initial condition for such a periodic orbit , and let @xmath58 be the flow at such initial condition .
then let us define the so - called ( subharmonic ) melnikov function @xmath59 then , if there exists @xmath60 such that 1 .
@xmath61 2 .
@xmath62 , then , for @xmath6 small enough , the perturbed system possesses a periodic orbit of period @xmath39 with initial condition @xmath63 at @xmath64 @xmath65 [ ex : analytical_melnikov ] compute the analytical expression of te melnikov function for system ( you do nt have to do the integral ! ) .
write a little program in matlab to numerically compute the melnikov integral obtained in exercise [ ex : analytical_melnikov ] .
we now want to numerically compute the initial conditions ( @xmath66 ) given by theorem [ theo : melnikov ] . here
we describe a newton method to do that in a general setting : @xmath34-dimensional not necessary hamiltonian systems .
assume we want to compute a periodic orbit of an @xmath34-dimensional non - autonomous periodic system of the form @xmath67 where @xmath68 is a @xmath8-periodic field in @xmath48 : @xmath69 , for any @xmath70 .
+ due to the periodicity , instead of using poincar maps in the state space it will be much more convenient to use the so - called stroboscopic map , which is indeed a poincar map but using a section in time .
this map is given by flowing system for a time @xmath8 with initial condition @xmath71 : @xmath72 provided that system is @xmath8-periodic in @xmath48 , @xmath73 becomes a map from the time section @xmath74 to itself : @xmath75 where @xmath41 imagine that we want to compute the initial condition , @xmath76 for a periodic orbit of system at the section @xmath77 .
recalling its periodicity , the period of such a periodic orbit must be a multiple of @xmath8 , say @xmath39 .
in other words , we must look for periodic points of the map @xmath78 , that is , points @xmath76 such that @xmath79 one of the most extended methods for numerically computing such points is the newton method , assuming that one has some idea about where such a points lies , as we need a first approximation for the newton method to converge where we want .
let us assume that we are looking for a @xmath8-periodic orbit ; that is , @xmath80 and we want to find a fixed point of the map @xmath78 .
then , we want to solve the equation @xmath81 the newton method consists of considering the linear approximation of the function @xmath82 around some point @xmath83 ( which is a first approximation of the solution we are looking for ) and solve the linear system instead .
this provides a new point , @xmath84 which , hopefully , is a more accurate solution than the initial one .
+ the linear approximation of equation @xmath85 around @xmath83 becomes @xmath86 where @xmath87 in section [ sec : variational ] we will see how to compute the differential @xmath88 .
+ if we solve equation for @xmath89 we get @xmath90 this leads to an iterative process @xmath91 which converges quadratically to @xmath76 provided that @xmath35 is a good enough approximation .
+ alternatively , some programing environments ( like matlab ) offer routines to solve linear equations which might be more efficient than computing the inverse @xmath92 . in this case , the linear system to be solved would be @xmath93 [ rem : df_invertibility ] note that the newton method requires the matrix @xmath94 to be invertible .
this implies two things : * @xmath95 must be invertible at the starting point * @xmath95 must be invertible at the fixed point !
this implies that the newton method will have troubles if the fixed point we are looking for is a center , as the eigenvalues of @xmath96 would have real part equal to one . in other words
, the periodic orbit has to hyperbolic .
similarly we can apply the newton method to solve the equation @xmath97 to get the same expression as in eq . .
however , in this case the computation of @xmath95 becomes now a bit more tricky . using that @xmath98
, we apply the chain rule to get @xmath99 se we need to evaluate the differential @xmath100 at the points @xmath101 for @xmath102 .
+ although there is nothing wrong with this approach from the theoretical point of view , in next section we will see a numerical method to compute @xmath100 which makes the computation of @xmath103 straightforward , with needing to multiply matrices ( see remark [ rem : varia_nperiodic ] below ) .
now the question arises , how do we compute @xmath96 ?
note that the flow @xmath104 is straightforward to differentiate with respect to @xmath48 , as one recovers the field @xmath105 , but we need to differentiate it with respect to the initial condition @xmath35 ! but we can do the following . applying the fundamental theorem of calculus and the definition of the flow , we can write @xmath106 if we now differentiate with respect to @xmath89 , we get @xmath107 where @xmath108 is the @xmath109 identity matrix and we write @xmath110 to emphasize that we differentiate with respect to @xmath89 . again , by applying the fundamental theorem of calculus backwards , we realize that equation is the solution of the differential equation @xmath111 at @xmath112 .
equation is called the ( first ) variational equation . + some remarks : if @xmath68 is a field in @xmath113 , then this equation becomes and @xmath109-dimensional differential equation .
equation is evaluated along the flow @xmath114 , which is unknown .
hence , this equation needs to be solved together with the equation @xmath115 , leading to a system of dimension @xmath116 with initial condition @xmath117 .
[ rem : varia_nperiodic ] if we want to compute @xmath103 , we just need to integrate the variational equations from @xmath118 to @xmath119 !
write a program in matlab in order to compute a fixed point of the stroboscopic map .
note that the initial sead will be taken from a zero of the melnikov function , which may have several .
you will need to tune @xmath45 to guarantee that this fixed point exists .
compute the eigenvalues of @xmath96 at the fixed points to tell their type .
as noted in remark [ rem : df_invertibility ] , the newton method to find fixed points ( or periodic orbits ) of the stroboscopic map will fail if that one is non - hyperbolic : that is , the associated eigenvalues of the stroboscopic map have real part equal to @xmath120 .
alternatively , we can use a poincar map using a section in the state space .
this method is more robust in that sense , but , as we will show below , the computation of the differential becomes slightly more tricky .
let us consider the poincar map @xmath121 where @xmath122{c } \varphi_1(t;x_0,t_0)\\\varphi_2(t;x_0,t_0 ) \end{array } \right ) \label{eq : perturbed_flow_notation}\ ] ] is the solution of the perturbed system with initial condition @xmath35 at @xmath74 and @xmath123 is such that .
it can be easily done using a newton method to solve equation .
below we show how to compute the necessary derivatives for the newton method .
otherwise , it can be computed using the `` events '' functionality of matlab . ]
@xmath124 for the second time , as we consider a return map in the direction that the section is left .
recall that system is non - autonomous .
hence , the initial time @xmath118 matters and we have to carry it on . from now on
, we will abuse notation in equation and omit writing the first coordinate @xmath125 , as it takes the @xmath126 . taking into account the periodicity of the non - autonomous system , an initial condition at @xmath127 for
@xmath74 of a periodic orbit of the full system will need to satisfy @xmath128{c } v_0\\t_0+mt \end{array } \right ) \label{eq : periodic_orbit_poinc - map}\ ] ] for some @xmath129 .
indeed , the integers @xmath34 and @xmath130 are the same as in previous sections , made explicit through the congruency equation .
here one clearly see that the role of @xmath34 is to count the `` loops '' that a periodic orbit of period @xmath39 makes around the origin .
applying the newton method to the equation @xmath131{c } v_0\\t_0+mt \end{array } \right)=\left ( \begin{array}[]{c } 0\\0 \end{array } \right ) , \label{eq : newton_equation_poinc - map}\ ] ] and arguing as in section [ sec : hyperbolic_po ] we get the iterative process @xmath132 where now @xmath133{cc } 1&0\\0&1 \end{array } \right).\ ] ] we now wonder , how do we compute @xmath134 ? let us see it for @xmath135 first .
+ recalling that @xmath136 , we get @xmath137{cc } d_{v_0}\varphi_2(t^*;0,v_0,t_0 ) & d_{t_0}\varphi_2(t^*;0,v_0,t_0)\\ \frac{\partial}{\partial v_0}t^ * & \frac{\partial}{\partial t_0}t^*\\ \end{array } \right).\ ] ] note that , in the first row , we have written the total derivatives @xmath138 and @xmath139 instead of partial ones , @xmath140 and @xmath141 , because @xmath123 actually depends on @xmath142 and @xmath118 through equation .
let us compute such total derivatives .
+ for the first one we get @xmath143 let us now see how to compute all the terms appearing in this equation .
+ on one hand , @xmath144 where @xmath145 is the second coordinate of the field evaluated at the image of the poincar map .
+ on the other one , @xmath146 is given by integrating the variational equations from @xmath118 to @xmath123 , as we are differentiating with respect an initial condition .
+ what about @xmath147 ? recall that , for given @xmath142 and @xmath118 , @xmath123 ( which we somehow know how to compute ) solves equation .
therefore , assuming that @xmath148 we can use the implicit function theorem to get @xmath149 note that , condition is satisfied , as the flow is transversal to the section @xmath27 .
if it were tangent , then we would have a problem , of course !
+ the denominator of the last equation is just the first coordinate of the perturbed field evaluated at the image of the poincar map .
+ regarding @xmath150 , note that it is a derivative with respect to the initial time , @xmath118 .
adding time as a variable , this can be computed integrating the variational equations of the system @xmath151 where now @xmath78 plays the role of time and @xmath118 becomes @xmath152 .
+ proceeding similarly , the elements of the second row of @xmath153 become @xmath154 and @xmath155 is already given in equation + the differential @xmath156 can be computed proceeding similarly as in section [ sec : differential_po ] , mutliplying @xmath153 evaluated at the iterates @xmath157 or integrating the previous variational equations until the @xmath158-th impact with the section @xmath27 occurs .
write a program in matlab to compute initial conditions for periodic orbits by performing a newton method to equation .
for that you will need to slightly modify the programs you wrote in the previous exercises . | these notes were written during the 9th and 10th sessions of the subject dynamical systems ii coursed at dtu ( denmark ) during the winter semester 2015 - 2016 .
they aim to provide students with a theoretical and numerical background for the computation of periodic orbits using newton s method .
we focus on periodically perturbed quasi - integrable systems ( using the forced pendulum as an example ) and hence we take advantage of the melnikov method to get first guesses . however , these well known techniques are general enough to be applied in other type of systems .
periodic orbits are computed by solving a fixed - point equation for the stroboscopic map , which is very fast and precise for hyperbolic periodic orbits .
however , for non - hyperbolic ones the method fails and we use the poincar map instead . in both cases
we show how to compute the jacobian of the maps , which is necessary for the newton method , by means of variational equations and the implicit function theorem . + some exercises are proposed along the notes , whose solutions can be found in github.com/a-granados .
+ the notes themselves do not contain any reference , although everything described here is well known in the dynamical systems community .
a typical reference for the melnikov method for subharmonic orbits is the book @xcite .
more about the variational equations and their numerical applications can be found in the notes @xcite . |
in our previous work @xcite , it was shown that in the strong coupling limit , when the number of flavors @xmath0 is greater than or equal to 7 , quarks are deconfined and chiral symmetry is restored at zero temperature , if the quark mass is lighter than a critical value . on the other hand , it is well - known that when @xmath0 exceeds @xmath5 , asymptotic freedom is lost . in this work
we investigate the problem of what is the condition on the number of flavors for quark confinement and spontaneous chiral symmetry breaking in the continuum limit , using the wilson quark action and the standard one - plaquette gauge action .
we first clarify the phase structure at zero temperature for general number of flavors , in particular , for @xmath6 .
when the phase diagram becomes clear , we are able to see what kind of continuum limit exists and eventually answer the problem given above . in order to reveal the structure of the deconfining phase , we increase
@xmath0 up to 300 , because the region of the deconfining phase at zero temperature becomes larger with larger @xmath0 , and consequently the structure becomes clear , as shown below .
then combining all data including the case of small @xmath0 , we are able to conjecture the phase structure for general number of flavors . through this systematic study of the deconfining phase for general number of flavors ,
we are also able to understand the behavior of the quark mass @xmath7 ( defined through an axial - vector ward identity ) and @xmath8 as functions of @xmath9 in the deconfining phase at finite temperatures in particular at small @xmath10 for small number of flavors such as @xmath4 and 3 .
the lattice sizes are @xmath11 ( @xmath12 , 6 or 8) , @xmath13 ( @xmath14 ) and @xmath15 ( @xmath16 ) .
we use an anti - periodic boundary condition for quarks in the @xmath17 direction and periodic boundary conditions otherwise .
when the hadron spectrum is calculated , the lattice is duplicated in the direction of lattice size 10 for @xmath18 , which we call the @xmath19 direction .
we call the pion screening mass simply the pion mass and similarly for the quark mass
. the current quark mass @xmath7 is defined by an axial - vector ward identity @xcite .
we use the hybrid r algorithm for the quark matrix inversion . as @xmath0 increases we have to decrease @xmath20 , such as @xmath20 = 0.0025 for @xmath21 , to reduce @xmath22 errors .
we have checked that the errors are sufficiently small selecting typical cases .
we first investigate the free case , because the free case is important as a reference to understand the structure of the deconfining phase .
we will show later that the @xmath9 dependence of the pion mass in the deconfining phase is quite similar to that of free quarks for any @xmath0 and @xmath10 . because the wilson formalism lifts the doublers by the wilson term
, there is only one pole in the free quark propagator at @xmath24 .
however , there is reminiscence of doublers . in fig .
[ free ] we plot the effective pion mass obtained at @xmath25 from the pion propagator .
the four cases shown there correspond to those for @xmath26 and 8 with an anti - periodic or the periodic boundary condition in the @xmath17 direction . in the case of the periodic boundary condition , there are almost no differences between the @xmath26 and 8 cases .
the vanishing of @xmath7 at @xmath27 and the behavior of @xmath8 for @xmath28 roughly given by @xmath29 are all as naively expected . for @xmath30 ,
the behavior of @xmath8 is somewhat complicated .
the vanishing of @xmath31 at @xmath32 is due to doublers ; one of momenta in spatial direction is @xmath33 . at @xmath34 ,
the momentum - energy ( dispersion ) relation is completely opposite to the usual one : the energy is a decreasing function of the magnitude of momentum for the case of equal momenta in three spatial dimensions .
the energy becomes infinity at zero momenta like the infinite mass case , while it is finite for other modes of momenta .
therefore the appearance of the peak of @xmath31 at @xmath34 is due to a superposition of contributions from several modes of momenta .
the behavior of @xmath7 defined through the axial - vector ward identity is also complicated at @xmath30 due to doublers . in the following ,
we refer to the rightmost @xmath35 state ( at @xmath27 for @xmath23 ) as simply the massless quark , or @xmath35 , distinguished from the other @xmath35 states , unless otherwise stated .
when an anti - periodic boundary condition is imposed , the position as well as the height of the peak of @xmath31 at @xmath34 does not change so much . on the other hand , for the massless quark at @xmath27 ,
the pion mass takes a value which is roughly twice the lowest matsubara frequency , for the both cases of @xmath36 and 8 .
the behavior around @xmath37 is similar , but indicates more complicated structure .
we now study the case of @xmath39 as in the previous work @xcite .
[ kd.nf]a shows the results of the deconfining transition points @xmath38 at @xmath39 obtained on @xmath36 lattices for various numbers of @xmath0 ( see also fig .
[ kd.nf]b ) . the data of plaquette ( fig .
[ nf240.pl ] ) for @xmath21 at @xmath39 indicates that the deconfining transition is of first order and that the location of the transition point is independent of @xmath40 for large @xmath40 ( @xmath41 for @xmath36 and @xmath42 for @xmath43 and 16 ) .
therefore we conclude that the transition is bulk as is confirmed in our previous work for @xmath44 @xcite . in fig .
[ kd.nf]a the transition points at @xmath45 for @xmath44 , 240 and 300 are also included .
these values are roughly those for the bulk transition points at zero temperature . as @xmath0 increases up to 240 , @xmath46 approaches toward or exceeds the value 8 , which is the value of @xmath9 for a massless free quark . because of this , for @xmath47 , we are able to see a wide region of the deconfining phase in the whole range of @xmath10
. therefore we first intensively investigate the case @xmath21 , and then decrease @xmath0 .
fig . [ nf240.mass ] shows the results of @xmath8 and @xmath48 for @xmath21 at @xmath39 , 2.0 , 4.5 , 6.0 , and 100 on the @xmath36 lattice .
a very striking fact is that the shape of @xmath8 and @xmath48 as a function of @xmath9 only slightly changes for @xmath49 when the value of @xmath10 decreases from @xmath50 down 0 .
only the position of the local minimum of @xmath8 at @xmath51 , which corresponds to the vanishing point of @xmath7 , slightly shifts toward smaller @xmath9 .
we obtain similar results also for @xmath43 . from these data we propose the phase diagram in fig .
[ nf240.phase ] for @xmath21 .
the dark shaded line is the phase boundary between the confining phase and the deconfining phase at zero temperature .
when @xmath12 or 8 , the boundary line bends down at finite @xmath10 as shown in fig .
[ nf240.phase ] , due to the finite temperature phase transition of the confining phase .
the dashed line corresponds to the @xmath35 line .
this line also corresponds to the minimum point of @xmath8 .
we have also calculated the quark propagator in the landau gauge and checked that chiral symmetry of the propagator , @xmath52 , is actually satisfied on the @xmath35 line .
the results for the case @xmath53 are essentially the same as those for @xmath21 except for very small shifts of the transition point and the minimum point of @xmath8 .
the @xmath35 point at @xmath23 is the trivial ir fixed point for @xmath1 .
the phase diagram shown in fig .
[ nf240.phase ] suggests that there are no other fixed points on the @xmath35 line at finite @xmath10 . in order to confirm this
, we investigate the direction of renormalization group ( rg ) flow along the @xmath35 line for @xmath21 , using a monte carlo renormalization group ( mcrg ) method .
one problem here is that it is practically impossible to make simulations in the massless limit at zero temperature due to the existence of zero modes in the quark matrix .
therefore , we impose an anti - periodic boundary condition in the @xmath17 direction .
we make a block transformation for a change of scale factor 2 , and estimate the quantity @xmath54 for the change of @xmath55 .
we generate configurations on an @xmath56 lattice on the @xmath35 points at @xmath39 and 6.0 and make twice blockings .
we also generate configurations on a @xmath57 lattice and make once a blocking .
then we calculate @xmath58 by matching the value of the plaquette at each step .
gauge theory , in particular in the deconfining phase , that one has to make a more careful analysis using several types of wilson loop to extract a precise value of @xmath54 .
we reserve elaboration of this point and a fine tuning of @xmath9 at each @xmath10 for future works . ] from the matching , we obtain @xmath59 for @xmath39 and @xmath60 for @xmath61 . the value obtained from the perturbation theory is @xmath62 at @xmath63 for @xmath21 .
the signs are the same and the magnitudes are comparable .
this suggests that the directions of rg flow on the @xmath35 line at @xmath39 and 6.0 are the same as that at @xmath23 for @xmath21 .
this further suggests that there are no fixed points at finite @xmath10 .
all of the above imply that the theory is trivial in the case of @xmath21 .
now we decrease the value of @xmath0 from 240 . fig .
[ 300to18 ] shows @xmath8 and @xmath48 for various @xmath0 , from 300 down to 18 , at @xmath61 and 0 .
when @xmath61 , the shapes of @xmath8 are almost identical to each other , except for a slight shift toward smaller @xmath9 as @xmath0 decreases .
when @xmath39 , the boundary of the first order phase transition between the deconfining phase and the confining phase moves toward smaller @xmath9 and therefore the range of the deconfining phase decreases : @xmath65 , 8 , 7.6 , 7.2 , 6.1 for @xmath66 , 240 , 180 , 120 and 60 , respectively , as shown in fig .
[ kd.nf ] . for @xmath67 ,
the value of @xmath46 decreases down to 5.0 . due to the fact that the confining phase invades the deconfining phase ,
the massless line in the deconfining phase hits the boundary at finite @xmath10 when @xmath0 becomes small . for example , in the case of @xmath67 , as shown in fig .
[ nf18.phase ] , it hits at @xmath68 4.5 .
although the area of the deconfining phase decreases with decreasing @xmath0 , the shape and the position of @xmath8 in the part of deconfining phase only slightly change from @xmath53 to 18 .
the values of @xmath7 as functions of @xmath9 also show only slight changes in the deconfining phase .
these facts , combined with the perturbative result that for @xmath1 , @xmath23 is the ir fixed point , suggest that the rg flow along the massless quark in the deconfining phase is the same as that for @xmath21 . in this case , the theory is trivial for @xmath1 .
case , we do not expect any qualitative differences between the cases of @xmath69 and 18 . ]
when @xmath70 , the theory is asymptotic free . on the other hand , quark confinement is lost for @xmath6 even in the strong coupling limit @xmath39 @xcite .
therefore the question is what happens for the cases @xmath2 in the deconfining phase .
we have intensively simulated the cases @xmath71 and 7 .
see the results for the case @xmath44 in fig .
the gross feature of the phase diagram does not much differ from the case of @xmath67 which is shown in fig .
[ nf18.phase ] .
one crucial point is that there is no chiral limit in the confining phase at least for @xmath72 .
this suggests that quarks are not confined in the continuum limit .
a more subtle point is whether there is a difference between the case @xmath1 and the case @xmath2 concerning the deconfining phase . for @xmath70 , the massless point at @xmath23 is a uv fixed point .
therefore the direction of the rg flow around @xmath23 is opposite to that for @xmath1 . in order to study the phase structure around @xmath73 ,
we have performed a mcrg study for the case @xmath71 similar to that for the case of @xmath21 .
we have investigated the rg flow along the maximum of @xmath8 , because this line roughly corresponds to an infinite quark mass line in a sense explained in sec .
[ sect : free ] .
we have found that the rg flow at @xmath39 and @xmath61 are opposite and the flow changes the direction around @xmath74 .
this result suggests that there is a fixed point on the @xmath35 line .
the flow of rg on the @xmath35 line at @xmath75 is the same as that at @xmath23 , which is opposite to that of the case of @xmath21 .
this implies that the location of the fixed point is at @xmath76 . because of the fact that the massless line @xmath35 hits the boundary between the two phases around @xmath74 , it is difficult to determine the location of the fixed point .
although we have not been able to locate the fixed point , we propose the phase structure , based on the consideration given above , that there is a non - trivial ir fixed point at finite @xmath10 @xcite for @xmath2 . of
course we have to make a detailed rg study to confirm that there is a non - trivial fixed point , which we reserve for future works .
the existence of a non - trivial fixed point implies that the theory in the continuum limit is a non - trivial theory with anomalous dimensions , however , without confinement .
for @xmath77 and 3 at finite temperatures , it was shown that @xmath7 has an unexpected @xmath9 dependence in the deconfining phase when @xmath10 is smaller than about 5.3 @xcite .
the @xmath8 as a function of @xmath9 also shows a cusp . in this section
we concentrate on these issues .
( for the discussion on the phase structure at zero temperature , see ref .
@xcite and references in it . ) = 6.9 cm we point out that the behaviors of @xmath7 and @xmath31 in the high temperature deconfining phase for @xmath3 at @xmath78 are similar to those in the deconfining phase at zero temperature for larger @xmath0 . in fig .
[ nf2.mq ] , the value of @xmath48 is plotted vs. @xmath9 for from @xmath21 down to @xmath4 at @xmath79 .
results at @xmath75 are similar .
the value of @xmath48 for the free quark case is also plotted .
there are two major trends when @xmath0 decreases from 240 down to 2 : one is an overall shift of the function @xmath48 in the deconfining phase to smaller @xmath9 and shrinkage of the shape with decreasing @xmath0 . the other is a shift of the transition point from the confining phase to the deconfining phase toward smaller @xmath9 . it should be noted that except for the overall shift and shrinkage mentioned above , the shape of the function @xmath48 in terms of @xmath9 in the deconfining phase is quite similar for all @xmath0 and also similar to that of a free quark .
thus the behavior of @xmath7 as well as @xmath8 as functions of @xmath9 in the deconfining phase at @xmath78 can be understood essentially as those of free quarks for @xmath30 .
= 6.6 cm we further note that for @xmath4 and 3 , the massless quark point ( the rightmost @xmath35 ) in the deconfining phase is different from the @xmath80 point in the confining phase for @xmath78 .
the difference is illustrated in the phase diagram for @xmath4 on the @xmath81 plane ( see fig . [ nf2.phase ] ) .
on the @xmath80 line , both @xmath8 and @xmath7 vanish in the confining phase at zero temperature @xcite .
the finite temperature transition for @xmath36 occur on the line @xmath82 .
the massless quark line for @xmath36 in the high temperature deconfining phase agrees with the @xmath80 line at @xmath83 .
however , they deviate from each other at @xmath78 .
this corresponds to the fact that the value of @xmath7 does not depend on the phase when @xmath84 , while it strongly depends on the phase when @xmath78 @xcite .
based on numerical results combined with the result of the perturbation theory we propose the following picture : there are three categories depending on the number of flavors @xmath0 : a free theory for @xmath1 , a non - trivial ir fixed point for @xmath2 and confinement and spontaneous chiral symmetry breaking for @xmath3 .
the conclusion that there are three categories depending on @xmath0 is in accord with recent theoretical works of @xmath85 super - symmetric qcd @xcite .
we also find that , for any @xmath6 and for @xmath3 at @xmath78 , the @xmath9 dependences of @xmath7 and @xmath8 in the deconfining phase are essentially the same as those of free quarks for @xmath30 , which show non - trivial structure due to doublers .
the peculiar behavior of @xmath7 as well as the cusp structure of @xmath8 in the deconfining phase at @xmath78 for @xmath4 @xcite can be understood in terms of the overall shift in @xmath9 and shrinkage of this structure .
numerical simulations are performed with hitac s820/80 at kek , and fujitsu vpp500/30 and qcdpax at the university of tsukuba .
this work is in part supported by the grants - in - aid of ministry of education , science and culture ( nos.07np0401 , 07640375 and 07640376 ) .
9 y. iwasaki , k. kanaya , s. sakai , and t. yoshi , phys .
lett . 69 ( 1992 ) 21 .
m. bochicchio _ et al .
_ , nucl .
b262 ( 1985 ) 331 .
s. itoh _ et al .
. phys .
b274 ( 1986 ) 33 .
t. banks and a. zaks , nucl .
b196 ( 1982 ) 173 . c. bernard
_ et al . _ ,
d49 ( 1994 ) 3574 . c. bernard
_ et al . _ ,
d46 ( 1992 ) 4741 ; t. blum _ et al .
_ , _ ibid .
_ d50 ( 1994 ) 3377
. y. iwasaki _ et al .
_ , nucl .
b ( proc . suppl . ) 42 ( 1995 ) 502 . | we investigate and elucidate the phase structure of qcd for general number of flavors @xmath0 with wilson quarks , varying @xmath0 from 2 up to 300 . based on numerical results combined with the result of the perturbation theory we propose the following picture : when @xmath1
, there is only a trivial fixed point and therefore the theory in the continuum limit is trivial . on the other hand ,
when @xmath2 , there is a non - trivial fixed point and therefore the theory is non - trivial with anomalous dimensions , however , without quark confinement .
theories which satisfy both quark confinement and spontaneous chiral symmetry breaking in the continuum limit exist only for @xmath3 .
we also discuss the structure of the deconfining phase at finite temperatures for the small number of flavors such as @xmath4 and 3 , through a systematic study of it for general number of flavors .
# 1 |
recent years have seen extensive progress on understanding the nature and dynamics of topological solitons @xcite , and in particular of skyrmions . for the su(2 ) skyrme system , minimal - energy skyrmions resemble polyhedral shells @xcite ; for example , the 3-skyrmion looks like a tetrahedron @xcite . on the other hand ,
the hopf - type solitons in the skyrme - faddeev system ( where the field takes values in the 2-sphere @xmath0 ) resemble closed loops , which may be linked or knotted @xcite ; for example , the 3-soliton in this system looks like a slightly - twisted circular loop . this paper describes a natural one - parameter family of generalized skyrme systems , which interpolates between the standard su(2 ) skyrme model and the skyrme - faddeev model .
its minimum - energy solutions interpolate between polyhedral skyrmions and string - like hopf solitons .
the simplest way to describe the family is as follows . in the su(2 ) skyrme model , the field takes values in the 3-sphere @xmath1 , with its standard metric .
this 3-sphere is fibred over @xmath0 ( the hopf fibration ) ; and instead of the standard metric on @xmath1 , we can use a metric for which distances along the ( one - dimensional ) fibres are scaled by a factor which we denote @xmath2 .
so @xmath3 gives the standard skyrme system , whereas @xmath4 corresponds to the target space being the quotient @xmath0 , namely the skyrme - faddeev system .
the global symmetry so(4 ) in the @xmath3 case is broken to u(2 ) when @xmath5 ; and this in turn means that the generalized skyrmion solutions for @xmath5 have less symmetry than those for @xmath3 .
the system can also be formulated in terms of a pair @xmath6 of complex scalars , and as such is related to condensed - matter systems in which there are two flavours of cooper pairs @xcite .
the parameter @xmath7 then appears , in particular , as the coefficient of a term @xmath8 , where @xmath9 is the current density . in two spatial dimensions , and without the fourth - order skyrme terms
, the case @xmath4 corresponds to the @xmath10 model .
the generalization of this to @xmath11 was investigated in @xcite .
it arises as a modification of the @xmath10 model which takes account of the effect of fermions ( starting with a system which has fermions as well as bosons , and integrating out the fermionic degrees of freedom ) . in this case
, there are explicit finite - energy static solutions ( parametrized by @xmath7 ) which , for @xmath4 , are the usual instanton solutions of the 2-dimensional @xmath10 model . in the three - dimensional case which is discussed below
, one needs a skyrme term to stabilize the solutions ; and the solutions have to be obtained numerically .
let us consider , first , the general situation of a map @xmath12 from a 3-space ( with local coordinates @xmath13 and metric @xmath14 ) to another 3-space ( with local coordinates @xmath15 and metric @xmath16 ) . the skyrme energy density @xmath17 of such a map may be defined as follows @xcite , in terms of the differential @xmath18 of @xmath12 .
define a @xmath19 matrix @xmath20 by @xmath21 then @xmath22 , where @xmath23.\ ] ] here @xmath24 and @xmath25 are constants .
if the metric @xmath16 admits a group of symmetries ( isometries ) , then these will correspond to ( global ) symmetries of the system . in
what follows , we take the target space to be the 3-sphere @xmath1 equipped with a one - parameter family of u(2)-invariant metrics .
a particular member of this family is the standard so(4)-invariant metric , and the corresponding system is the usual su(2 ) skyrme model .
the family of metrics may be described as follows .
let @xmath6 denote a complex 2-vector satisfying the constraint @xmath26 ( where @xmath27 is the complex - conjugate row vector corresponding to the column vector @xmath28 ) .
the set of all such vectors @xmath28 forms a 3-sphere .
note that the map @xmath29 is the standard hopf fibration from @xmath1 to @xmath0 , with @xmath30 being the usual stereographic coordinate on @xmath0 .
the standard metric @xmath31 on @xmath1 corresponds to @xmath32 let @xmath33 be the vector field obtained from the 1-form @xmath34 by raising its index with the metric ( [ stmet ] ) .
this vector field @xmath33 has unit length , and is tangent to the fibres of the hopf fibration .
our family of metrics , parametrized by the real number @xmath7 , is taken to be @xmath35 .
an alternative way to write @xmath36 is @xmath37 note that both @xmath31 and @xmath38 , and hence also @xmath36 , are manifestly invariant under the u(2 ) transformations @xmath39 , where @xmath40 . for @xmath11 , the metric ( [ met ] )
is positive - definite .
but when @xmath4 , it becomes degenerate , with @xmath33 being a zero - eigenvector ; distances along the hopf fibres are then zero , and the metric is , in effect , the standard metric on the quotient space @xmath41 . in other words ,
our one - parameter family includes the standard 3-sphere ( @xmath3 ) and the standard 2-sphere ( @xmath4 ) .
we will restrict to the range @xmath42 for which the metric is non - negative ; in fact , our interest is in the range @xmath43 , which interpolates between the skyrme and the skyrme - faddeev systems .
the lagrangian @xmath44 of the generalized skyrme system ( consistent with the expressions ( [ gene ] ) for the static energy density ) may be described as follows .
the vector @xmath28 determines an su(2 ) matrix according to @xmath45.\ ] ] write @xmath46 = k_{\mu\nu } = { { \rm i}}k^a_{\mu\nu}\sigma_a,\ ] ] where the partial derivative is with respect to space - time coordinates @xmath47 , and where @xmath48 denotes the pauli matrices .
then @xmath49 , where @xmath50.\end{aligned}\ ] ] in this form , the global u(2 ) symmetry corresponds to @xmath51 , where @xmath52 is an su(2 ) matrix and @xmath53 is a diagonal su(2 ) matrix ; note that this transformation preserves both @xmath54 and @xmath55 . if @xmath3 , then @xmath44 is the standard skyrme lagrangian . if @xmath4 , on the other hand , we get the skyrme - faddeev system @xcite .
one way of seeing this is to replace the field @xmath28 by the unit 3-vector field @xmath56
. then @xmath44 with @xmath4 becomes @xmath57 where @xmath58 ; this is the skyrme - faddeev lagrangian .
if we take the space on which the field @xmath59 is defined to be @xmath60 , then we need a boundary condition @xmath61 ( constant ) as @xmath62 , to have finite energy .
fields satisfying this condition are classified topologically by their winding number @xmath63 , where @xmath64 is the topological charge density @xmath65 in the limit @xmath66 , @xmath67 equals the hopf number of the @xmath0-valued field . the values of the constants @xmath24 and @xmath25 correspond to the energy and length scales .
to choose convenient values for them in what follows , let us consider the system defined on the unit 3-sphere @xmath1 ( that is , take @xmath14 to be the standard metric on @xmath1 ) @xcite ; and take the field @xmath68 to correspond to the identity map from @xmath1 to itself ( in other words , an isometry if @xmath3 ) .
it is straightforward to compute the energy @xmath69 of this field : one gets @xmath70 so from now on let us take @xmath71 , \quad \lambda_4=1/[4\pi^2(3 - 2\alpha)];\ ] ] consequently , the ` identity ' field has unit energy for all @xmath72 $ ] .
a numerical minimization procedure was used to find local minima of the static energy @xmath69 for various values of @xmath67 and @xmath7 , and hence stable skyrmion solutions ; the results are described below . the procedure uses a finite - difference version of the functional @xmath69 on a cubic grid , with a second - order scheme in which the truncation error is of order @xmath73 where @xmath74 is the lattice spacing , and using the coordinate @xmath75 for @xmath76 ( similarly for @xmath77 and @xmath78 ) so that the whole of @xmath60 is included . with a relatively small number of lattice points ( say @xmath79 ) , this achieves an accuracy of better than @xmath80 .
the discrete energy was then minimized using a standard conjugate - gradient method ( flowing down the energy gradient ) .
this produces a local minimum of the energy functional . in general , there are many local minima ; the starting configuration determines which one is produced by this procedure .
it seems likely that the solutions described below are global minima in the relevant topological classes , but the only evidence for this at present is consistency with previous studies in the @xmath3 and @xmath4 cases @xcite .
most straightforward are the @xmath81 and @xmath82 cases , where the solutions admit a continuous symmetry . for @xmath81 ,
the @xmath3 skyrmion has o(3 ) ( spherical ) symmetry , and energy @xmath83 . when @xmath5 , this is broken to o(2 ) ( axial ) symmetry .
the normalized energy @xmath84 depends smoothly on @xmath7 , and the numerical results indicate that , to within the small numerical error , its dependence is quadratic : @xmath85 .
the topological charge density ( [ b ] ) has an almost - spherical shape , for all @xmath7 . for @xmath82 , one has o(2 ) symmetry both for @xmath3 and @xmath4 , and the constant-@xmath64 surfaces resemble tori .
so the expectation is that the @xmath82 generalized skyrmions will look like tori for all values of @xmath7 , with @xmath84 decreasing from @xmath86 to @xmath87 @xcite over the range @xmath72 $ ] ; but this has not been checked .
it is worth remarking at this point on the energy - values of skyrme - faddeev solitons given in @xcite , so as to facilitate comparison with that paper .
the energies in @xcite should be divided by a factor of @xmath88 in order to adjust the normalization to the one being used here ; and by a further factor of ( about ) @xmath89 to allow for the fact that @xcite used a finite - size box ( rather than all of @xmath60 ) .
for example , in the @xmath82 case , @xcite gives an energy @xmath90 , which when divided by the two factors above yields @xmath91 .
this is within @xmath92 of the correct figure . for @xmath96 ,
the picture is less straightforward , with the skyrmions having at most discrete symmetry .
we look in detail at the cases @xmath93 and @xmath97 .
the 3-skyrmion ( for @xmath3 ) has energy @xmath98 and tetrahedral symmetry @xcite ; in particular , a typical constant-@xmath64 surface resembles a tetrahedron .
it is also useful to plot the curve in @xmath60 where @xmath99 , or equivalently where @xmath100 and @xmath101 ; in the faddeev - skyrme system , this curve may be interpreted as the position of the string - like hopf skyrmion @xcite .
each plot in figure 1 depicts the surface @xmath102 , with the ` thickened ' curve @xmath103 strung around it ; subfigure ( a ) is for @xmath3 , ( b ) is for @xmath94 , and ( c ) is for @xmath95 .
we see that as @xmath7 increases from zero , the tetrahedral skyrmion transforms into a twisted torus or loop ( see also the pictures in @xcite for the @xmath4 case ) .
the tetrahedral symmetry is broken to the subgroup @xmath104 .
the normalized energy @xmath84 again has a quadratic dependence on @xmath7 : @xmath105 .
finally , let us look at the case @xmath97 .
the 4-skyrmion ( for @xmath3 ) resembles a cube @xcite : see figure 2(a ) , where the same quantities are plotted as in figure 1 . as @xmath7 increases , the minimum - energy configuration becomes a closed loop strung along eight edges of the cube ( figure 2(b ) , for @xmath106 ) , which then flattens as @xmath7 increases further .
when @xmath4 , one again gets a twisted circular loop , with the twisting being greater than in the @xmath93 case ( see also the pictures in @xcite ) .
we have seen that the skyrme model and the skyrme - faddeev - hopf system may be regarded as members of a one - parameter family of generalized skyrme systems ; and the topological - soliton solutions of all these systems , although rather different in appearance , are all closely - related to one another .
a recent paper @xcite has pointed out a similarity between sphaleron solutions of the skyrme system and axially - symmetric hopf solitons , especially as the winding number @xmath67 increases .
these solutions are unstable ( saddle - points of their respective energy functionals ) ; and this connection between skyrmions and hopf solitons is quite different from the one described above
. it may be of interest , however , to investigate sphaleron - type solutions of the family of skyrme systems , and see how they depend on the family parameter @xmath7 . * note added in proof . *
a similar family arises from considering bundles of strings [ s. nasir and a. j. niemi , mod .
lett . a 17 , 1445 ( 2002 ) ] .
the author is grateful to prof .
niemi for correspondence regarding this . | this paper describes a natural one - parameter family of generalized skyrme systems , which includes the usual su(2 ) skyrme model and the skyrme - faddeev system .
ordinary skyrmions resemble polyhedral shells , whereas the hopf - type solutions of the skyrme - faddeev model look like closed loops , possibly linked or knotted . by looking at the minimal - energy solutions in various topological classes , and for various values of the parameter
, we see how the polyhedral skyrmions deform into loop - like hopf skyrmions . |
a gp @xmath1 is a probabilistic map parametrised by a covariance @xmath2 and a mean @xmath3 .
we use @xmath4 and automatic relevance determination ( ard ) @xmath5 in the following . here ,
@xmath6 and @xmath7 denote the signal and noise variance , respectively and the diagonal matrix @xmath8 contains the squared length scales .
since a gp is a distribution over functions @xmath9 , the output @xmath10 is random even though the input @xmath11 is deterministic . in gp regression , a gp prior
is combined with training data @xmath12 into a gp posterior conditioned on the training data with mean @xmath13 and covariance @xmath14 where @xmath15@xmath16^{\top}$ ] , @xmath17_{ij=1 .. n}$ ] , @xmath18_{ij=1 .. n}$ ] and @xmath19\mathbf{k}^{-1}$ ] .
deterministic inputs @xmath11 lead to gaussian outputs and gaussian inputs lead to non - gaussian outputs whose first two moments can be computed analytically @xcite for ard covariance .
multivariate deterministic inputs @xmath11 lead to spherical gaussian outputs @xmath20 and gaussian inputs @xmath11 lead to non - gaussian outputs @xmath20 whose moments @xmath21 are given by : here , @xmath22^{\top}=[\mathbf{z}^{1}, .. ,\mathbf{z}^{n}]\mathbf{k}^{-1}$ ] and the quantities @xmath23 $ ] and @xmath24 $ ] denote expectations of @xmath25^{\top}$ ] w.r.t .
the gaussian input distribution @xmath26 that can readily be evaluated in closed form @xcite as detailed in the appendix . in the limit of @xmath27
we recover the deterministic case as @xmath28 , @xmath29 and @xmath30 .
non - zero input variance @xmath31 results in full non - spherical output covariance @xmath32 , even for independent gps because all the gps are driven by the same ( uncertain ) input .
a gplvm @xcite is a successful and popular non - parametric bayesian tool for high dimensional nonlinear data modeling taking into account the data s manifold structure based on a low - dimensional representation .
high dimensional data points @xmath33 , @xmath34 $ ] , are represented by corresponding latent points @xmath35 $ ] from a low - dimensional latent space @xmath36 mapped into @xmath37 by @xmath38 independent gps @xmath39
one for each component @xmath40 of the data .
all the gps @xmath39 are conditioned on @xmath41 and share the same covariance and mean functions .
the model is trained by maximising the sum of the log marginal likelihoods over the @xmath38 independent regression problems with respect to the latent points @xmath41 .
the high dimensional ( @xmath42 ) density of the data points @xmath43 in panel ( c ) is modelled by a mixture of gaussians @xmath44 shown in panel ( b , d ) where the means @xmath45 and variances @xmath46 are given by the predictive means and covariances of a set of independent gaussian processes @xmath47 conditioned on low - dimensional ( @xmath48 ) latent locations @xmath49 . a latent dirac mixture ( a )
yields a spherical gaussian mixture with varying widths ( b ) and a latent gaussian mixture ( e ) results in a fully coupled mixture model ( d ) smoothly sharing covariances across mixture components.,scaledwidth=100.0% ] while most often applied to nonlinear dimensionality reduction , the gplvm can also be used as a tractable and flexible density model in high dimensional spaces as illustrated in figure [ fig : dgplvm ] .
the basic idea is to interpret the latent points @xmath41 as centres of a mixture of either dirac ( figure [ fig : dgplvm]a ) or gaussian ( figure [ fig : dgplvm]e ) distributions in the latent space @xmath50 that are _ projected forward _ by the gp to produce a high dimensional gaussian mixture @xmath44 in the observed space @xmath37 .
depending on the kind of latent mixture , the density model @xmath51 will either be a mixture of spherical ( figure [ fig : dgplvm]b ) or full - covariance gaussians ( figure [ fig : dgplvm]d ) . by that mechanism
, we get a tractable high dimensional density model @xmath51 : a set of low - dimensional coordinates in conjunction with a probabilistic map @xmath1 yield a mixture of high dimensional gaussians whose covariance matrices are smoothly shared between components .
as shown in figure [ fig : dgplvm](d ) , the model is able to capture high dimensional covariance structure along the data manifold by relatively few parameters ( compared to @xmath0 ) , namely the latent coordinates @xmath52 and the hyperparameters @xmath53\in\mathbb{r}_{+}^{2d+2}$ ] of the gp .
the role of the latent coordinates @xmath41 is twofold : they both _ define the gp _ , mapping the latent points into the observed space , and they _ serve as centres _ of the mixture density in the latent space .
if the latent density is a mixture of gaussians , the centres of these gaussians are used to define the gp map , but the full gaussians ( with covariance @xmath31 ) are projected forward by the gp map .
learning or model fitting is done by minimising a loss function @xmath54 w.r.t .
the latent coordinates @xmath41 and the hyperparameters @xmath55 . in the following
, we will discuss the usual gplvm objective function , make clear that it is not suited for density estimation and use leave - out estimation to avoid overfitting .
a gplvm @xcite is trained by setting the latent coordinates @xmath41 and the hyperparameters @xmath55 to maximise the probability of the data @xmath56 + that is the product of the marginal likelihoods of @xmath38 independent regression problems . using @xmath57 , conjugate gradients optimisation at a cost of @xmath58 per step is straightforward but suffers from local optima .
however , optimisation of @xmath59 does not encourage the gplvm to be a good density model . only indirectly
, we expect the predictive variance to be small ( implying high density ) in regions supported by many data points .
the main focus of @xmath59 is on faithfully predicting @xmath60 from @xmath41 ( as implemented by the fidelity trace term ) while using a relatively smooth function ( as favoured by the log determinant term ) .
therefore , we propose a different cost function . density estimation @xcite constructs parametrised estimators @xmath61 from iid data @xmath62 .
we use the kullback - leibler divergence @xmath63 to the underlying density and its empirical estimate @xmath64 as quality measure where @xmath65 emphasises that the full dataset has been used for training .
this estimator , is prone to overfitting if used to adjust the parameters via @xmath66 .
therefore , estimators based on @xmath67 subsets of the data @xmath68 are used .
two well known instances are @xmath67-fold cross - validation ( cv ) and leave - one - out ( loo ) estimation .
the subsets for cv are @xmath69 and @xmath70 for loo .
both of them can be used to optimise @xmath55 .
there are two reasons why training a gplvm with the log likelihood of the data @xmath59 ( eq . [ eq : gplvm - lik ] ) is not optimal in the setting of density estimation : firstly , it treats the task as regression , and does nt explicitly worry about how the density is spread in the observation space .
secondly , our empirical results ( see section [ sec : experiments ] ) indicate , that the test set performance is simply not good .
therefore , we propose to train the model using the leave - out density @xmath71 + this objective is very different from the gplvm criterion as it measures how well a data point is explained under the mixture models resulting from projecting each of the latent mixture components forward ; the leave - out aspect enforces that the point @xmath43 gets assigned a high density even though the mixture component @xmath72 has been removed from the mixture .
the leave - one - out idea is trivial to apply in a mixture setting by just removing the contribution in the sum over components , and is motivated by the desire to avoid overfitting .
evaluation of @xmath73 requires @xmath58 assuming @xmath74 .
however , removing the mixture component is not enough since the latent point @xmath49 is still present in the gp . using rank
one updates to compute inverses and determinants of covariance matrices @xmath75 with row and column @xmath76 removed , it is possible to evaluate eq .
[ eq : dgplvm - lpo ] for mixture components @xmath77 with latent point @xmath49 removed from the mean prediction @xmath78 which is what we do in the experiments .
unfortunately , going further by removing @xmath49 also from the covariance @xmath79 increases the computational burden to @xmath80 because we need to compute rank one corrections to all matrices @xmath81 .
since @xmath82 is only slightly smaller than @xmath79 , we refrain from computing it in the experiments . in the original gplvm
, there is a clear one - to - one relationship between latent points @xmath49 and data points @xmath43 they are inextricably tied together .
however , the leave - one - out ( loo ) density @xmath73 does not impose any constraint of that sort . the number of mixture components does not need to be @xmath83 , in fact we can choose any number we like .
only the data visible to the gp @xmath84 is tied together .
the actual latent mixture centres @xmath41 are not necessarily in correspondence with any actual data point @xmath43 .
however , we can choose @xmath85 to be a subset of @xmath60 .
this is reasonable because any mixture centre @xmath86 ( corresponding to the latent centre @xmath87 ) lies in the span of @xmath85 , hence @xmath85 should approximately span @xmath60 . in our experiments , we enforce @xmath88 .
overfitting in density estimation means that very high densities are assigned to training points , whereas very low densities remain for the test points . despite its success in parametric models , the leave - one - out idea alone , is not sufficient to prevent overfitting in our model . when optimising @xmath73 w.r.t .
@xmath89 using conjugate gradients , we observe the following behaviour : the model circumvents the loo objective by arranging the latent centres in pairs that take care of each other . more generally , the model partitions the data @xmath90 into groups of points lying in a subspace of dimension @xmath91 and adjusts @xmath89 such that it produces a gaussian with very small variance @xmath92 in the orthogonal complement of that subspace .
by scaling @xmath92 to tiny values , @xmath73 can be made almost arbitrarily large .
it is understood that the hyperparameters of the underlying gp take very extreme values : the noise variance @xmath7 and some length scales @xmath93 become tiny . in @xmath59
, this is penalised by the @xmath94 term , but @xmath73 is happy with very improbable gps . in our initial experiments , we observed this `` cheating behaviour '' on several of datasets .
we conclude that even though the loo objective ( eq . [ eq : dgplvm - lpo ] ) is the standard tool to set kde kernel widths @xcite , it breaks down for too complex models .
we counterbalance this behaviour by leaving out not only one point but rather @xmath95 points at a time .
this renders cheating tremendously difficult . in our experiments we use the leave-@xmath95-out ( lpo ) objective @xmath96 + ideally , one would sum over all @xmath97 subsets @xmath98 of size @xmath99 .
however , the number of terms @xmath67 soon becomes huge : @xmath100 for @xmath101 .
therefore , we use an approximation where we set @xmath102 and @xmath103 contains the indices @xmath104 that currently have the smallest value @xmath105 . all gradients @xmath106 and @xmath107 can be computed in @xmath58 when using @xmath78 .
however , the expressions take several pages .
we use a conjugate gradient optimiser to find the best parameters @xmath41 and @xmath55 .
in the experimental section , we show that the gplvm trained with @xmath59 ( eq . [ eq : gplvm - lik ] ) does not lead to a good density model in general . using our @xmath108 training procedure ( section [ sub : overfit ] ,
[ eq : dgplvm - lpo ] ) , we can turn it into a competitive density model .
we demonstrate that a latent variance @xmath109 improves the results even further in some cases and that on some datasets , our density model training procedure performs better than all the baselines .
we consider @xmath110 data sets , frequently used in machine learning .
the data sets differ in their domain of application , their dimension @xmath38 , their number of instances @xmath83 and come from regression and classification . in our experiments
, we do not use the labels .
we do not only want to demonstrate that our training procedure yields better test densities for the gplvm .
we are rather interested in a fair assessment of how competitive the gplvm is in density estimation compared to other techniques . as baseline methods
, we concentrate on three standard algorithms : penalised fitting of a mixture of full gaussians ( ` gm ` ) , kernel density estimation ( ` kde ` ) and manifold parzen windows @xcite ` ( mp ) ` .
we run these algorithms for three different type of preprocessing : raw data ( r ) , data scaled to unit variance ( ` s ` ) and whitened data ( ` w ` ) . we explored a large number of parameter settings and report the best results in table [ tab : baselines ] . in order to speed up em computations , we partition the dataset into @xmath67 disjoint subsets using the @xmath67-means algorithm `` .
we fitted a penalised gaussian to each subset and combined them using the relative cluster size as weight @xmath111 .
every single gaussian @xmath112 has the form @xmath113 where @xmath114 and @xmath115 equal the sample mean and covariance of the particular cluster , respectively .
the global ridge parameter @xmath116 prevents singular covariances and is chosen to maximise the loo log density @xmath117 .
we use simple gradient descent to find the best parameter @xmath118 .
the kernel density estimation procedure fits a mixture model by centring one mixture component at each data point @xmath43 .
we use independent multi - variate gaussians : @xmath119 , where the diagonal widths @xmath120 are chosen to maximise the loo density @xmath121 .
we employ a newton - scheme to find the best parameters @xmath122 .
the manifold parzen window estimator @xcite tries to capture locality by means of a kernel @xmath123 .
it is a mixture of @xmath83 full gaussians where the covariance @xmath124 of each mixture component is only computed based on neighbouring data points .
as proposed by the authors , we use the @xmath125-nearest neighbour kernel and do not store full covariance matrices @xmath126 but a low rank approximation @xmath127 with @xmath128 . as in the other baselines ,
the ridge parameter @xmath116 is set to maximise the loo density .
the results of the baseline density estimators can be found in table [ tab : baselines ] .
they clearly show three things : ( i ) more data yields better performance , ( ii ) penalised mixture of gaussians is clearly and consistently the best method and ( iii ) manifold parzen windows @xcite offer only little benefit .
the absolute values can only be compared within datasets since linearly transforming the data @xmath60 by @xmath129 results in a constant offset @xmath130 in the log test probabilities .
we keep the experimental schedule and setting of the previous section in terms of the @xmath110 datasets , the @xmath131 fold averaging procedure and the maximal training set size @xmath132 .
we use the gplvm log likelihood of the data @xmath59 , the lpo log density with deterministic latent centres ( @xmath133 ) and the lpo log density using a gaussian latent centres @xmath134 to optimise the latent centres @xmath41 and the hyperparameters @xmath55 .
our numerical results include @xmath135 different latent dimensions @xmath136 , @xmath135 preprocessing procedures and @xmath137 different numbers of leave - out points @xmath95 . optimisation is done using @xmath138 conjugate gradient steps alternating between @xmath41 and @xmath55 . in order to compress the big amount of numbers ,
we report the method with highest test density as shown in figure [ fig : test_density ] , only .
each panel displays the log test density averaged over @xmath131 random splits for three different gplvm training procedures and the _ best _ out of 41 baselines ( penalised mixture @xmath139 , diag.+isotropic kde , manifold parzen windows with 36 different parameter settings ) as well as various mixture of factor analysers ( mfa ) settings as a function of the number of training data points @xmath140 . we report the maximum value across latent dimension @xmath141 , three preprocessing methods ( raw , scaled to unit variance , whitened ) and @xmath142 leave - out points . the gplvm training procedures are the following : @xmath108-rd : stochastic leave-@xmath95-out density ( eq . [ eq : dgplvm - lpo ] with latent gaussians , @xmath109 ) , @xmath108-det : deterministic leave-@xmath95-out density ( eq . [ eq : dgplvm - lpo ] with latent diracs , @xmath143 ) and @xmath144 : marginal likelihood ( eq . [ eq : gplvm - lik]).,scaledwidth=100.0% ] the most obvious conclusion , we can draw from the numerical experiments , is the bad performance of @xmath59 as a training procedure for gplvm in the context of density modeling .
this finding is consistent over all datasets and numbers of training points .
we get another conclusive result in terms of how the latent variance @xmath31 influences the final test densities could be fixed to @xmath145 because its scale can be modelled by @xmath41 . ] . only in the ` bodyfat
` data set it is not beneficial to allow for latent variance .
it is clear that this is an intrinsic property of the dataset itself , whether it prefers to be modelled by a spherical gaussian mixture or by a full gaussian mixture .
an important issue , namely how well a fancy density model performs compared to very simple models , has in the literature either been ignored @xcite or only done in a very limited way @xcite . experimentally
, we can conclude that on some datasets e.g. ` diabetes , sonar , abalone ` our procedure can not compete with a plain ` gm ` model .
however note , that the baseline numbers were obtained as the maximum over a wide ( 41 in total ) range of parameters and methods .
for example , in the ` usps ` case , our elaborate density estimation procedure outperforms a single penalised gaussian only for training set sizes @xmath146 .
however , the margin in terms of density is quite big : on @xmath147 prewhitened data points @xmath148 with deterministic latents yields @xmath149 at @xmath150 , whereas full @xmath148 reaches @xmath151 at @xmath152 which is significantly above @xmath153 as obtained by the ` gm ` method since we work on a logarithmic scale , this corresponds to factor of @xmath154 in terms of density . while the baseline methods such as ` gm , ` ` kde ` and ` mp ` run in a couple of minutes for the ` usps ` dataset , training a gplvm with either @xmath155 or @xmath15@xmath59 takes considerably longer since a lot of cubic covariance matrix operations need to be computed during the joint optimisation of @xmath89 .
the gplvm computations scale cubically in the number of data points @xmath140 used by the gp forward map and quadratically in the dimension of the observed space @xmath38 .
the major computational gap is the transition from @xmath143 to @xmath109 because in the latter case , covariance matrices of size @xmath0 have to be evaluated which cause the optimisation to last in the order of a couple of hours . to provide concrete timing results , we picked @xmath147 , @xmath150 , averaged over the @xmath110 datasets and show times relative to @xmath144 .
note that the methods @xmath108 are run in a conservative fail - proof black box mode with @xmath138 gradient steps .
we observe good densities after considerably less gradient steps already .
another straightforward speedup can be obtained by carefully pruning the number of inputs to the @xmath108 models .
we have discussed how the basic gplvm is not in itself a good density model , and results on several datasets have shown , that it does not generalise well .
we have discussed two alternatives based on explicitly projecting forward a mixture model from the latent space .
experiments show that such density models are generally superior to the simple gplvm . among the two alternative ways of defining the latent densities ,
the simplest is a mixture of delta functions , which due to the stochasticity of the gp map results in a smooth predictive distribution .
however , the resulting mixture of gaussians , has only axis aligned components .
if instead the latent distribution is a mixture of gaussians , the dimensions of the observations become correlated .
this allows the learnt densities to faithfully follow the underlying manifold .
although the presented model has attractive properties , some problems remain : the learning algorithm needs a good initialisation and the computational demand of the method is considerable .
however , we have pointed out that in contrast to the gplvm , the number of latent points need not match the number of observations allowing for alternative sparse methods . | density modeling is notoriously difficult for high dimensional data .
one approach to the problem is to search for a lower dimensional manifold which captures the main characteristics of the data .
recently , the gaussian process latent variable model ( gplvm ) has successfully been used to find low dimensional manifolds in a variety of complex data .
the gplvm consists of a set of points in a low dimensional latent space , and a stochastic map to the observed space .
we show how it can be interpreted as a density model in the observed space .
however , the gplvm is not trained as a density model and therefore yields bad density estimates .
we propose a new training strategy and obtain improved generalisation performance and better density estimates in comparative evaluations on several benchmark data sets .
modeling of densities , aka unsupervised learning , is one of the central problems in machine learning . despite its long history @xcite ,
density modeling remains a challenging task especially in high dimensional spaces .
for example , the generative approach to classification requires density models for each class , and training such models well is generally considered more difficult than the alternative discriminative approach .
classical approaches to density modeling include both parametric and non parametric methods . in general ,
simple parametric approaches have limited utility , as the assumptions might be too restrictive .
mixture models , typically trained using the em algorithm , are more flexible , but e.g. gaussian mixture models are hard to fit in high dimensions , as each component is either diagonal or has in the order of @xmath0 parameters , although the mixtures of factor analyzers algorithm @xcite may be able to strike a good balance .
methods based on kernel density estimation @xcite are another approach , where bandwidths may be set using cross validation @xcite .
the methods mentioned so far have two main shortcomings : 1 ) they typically do not perform well in high dimensions , and 2 ) they do not provide an intuitive or generative understanding of the data .
generally , we can only succeed if the data has some regular structure , the model can discover and exploit .
one attempt to do this is to assume that the data points in the high dimensional space lie on or close to some smooth underlying lower dimensional manifold .
models based on this idea can be divided into models based on _ implicit _ or _ explicit _ representations of the manifold .
an implicit representation is used by @xcite in a non - parametric gaussian mixture with adaptive covariance to every data point .
explicit representations are used in the generative topographic map @xcite and by @xcite . within the explicit camp ,
models contain two separate parts , a lower dimensional latent space equipped with a density , and a function which maps points from the low dimensional latent space to the high dimensional space where the observations lie .
advantages of this type of model include the ability to understand the structure of the data in a more intuitive way using the latent representation , as well as the technical advantage that the density in the observed space is automatically properly normalised by construction . the gaussian process latent variable model ( gplvm ) @xcite uses a gaussian process ( gp ) @xcite to define a ( stochastic ) map between a lower dimensional latent space and the observation space .
however , the gplvm does not include a density in the latent space . in this paper
, we explore extensions to the gplvm based on densities in the latent space .
one might assume that this can trivially be done , by thinking of the latent points learnt by the gplvm as representing a mixture of delta functions in the latent space .
since the gp based map is stochastic , it induces a proper mixture in the observed space .
however , this formulation is unsatisfactory , because the resulting model is not trained as a density model .
consequently , our experiments show poor density estimation performance .
mixtures of gaussians form the basis of the vast majority of density estimation algorithms .
whereas kernel smoothing techniques can be seen as introducing a mixture component for each data point , infinite mixture models @xcite explore the limit as the number of components increases and mixtures of factor analysers impose constraints on the covariance of individual components . the algorithm presented in this paper
can be understood as a method for stitching together gaussian mixture components in a way reminiscent of @xcite using the gplvm map from the lower dimensional manifold to induce factor analysis like constraints in the observation space . in a nutshell
, we propose a density model in high dimensions by transforming a set of low - dimensional gaussians with a gp .
we begin by a short introduction to the gplvm and show how it can be used to define density models . in section [ sec : learning ] , we introduce a principled learning algorithm , and experimentally evaluate our approach in section [ sec : experiments ] . |
the study of localized surface plasmon ( lsp ) resonances over the last couple of decades has led to great advances in several areas of science and technology .
probably , its most significant application is its use in nanosensors.@xcite most of the strategies used to turn a plasmonic device into a sensor are based on the shift of the resonant position of the lsp@xcite or in exploiting hot spots to increase spectroscopic signals.@xcite in this respect , there is a growing interest in using interacting plasmonic systems to measure distances at the nanoscale .
these devices are known as nanorulers and they are finding interesting applications in several fields such as biology.@xcite most of the proposed plasmonic rulers work by the same principle , the shift of a resonance induced by the variation of a distance . as noticed originally by jain et .
al.,@xcite the frequency shift of the lsp resonance follows an almost universal exponential or quasi - exponential law with respect to the separation between nps .
this behavior is known as the universal plasmon ruler equation ( upre ) and was observed in simulations and experiments using nps of different size and shape,@xcite core - shell nps,@xcite and even two dimensional arrays of nps.@xcite there are some deviations of the upre due to retardation effects or the effects of higher order multipoles , but they can usually be corrected by empirical fittings.@xcite only when nps are almost touching each other , there are some deviations , nonfitable to exponential , of the upre.@xcite there is another kind of plasmon rulers that is now being actively investigated .
they are based on a different principle , the fluorescence resonance energy transfer ( fret ) .
its working principle is the resonant transference of excitation from a fluorescent molecule to a nearby np .
the closer to fluorophore to the np , the greater the photoluminescence quenching .
this effect has been successfully used as a nanorulers in several experimental examples.@xcite collective effects induced by lsp couplings in nps arrays can give rise to other phenomena and new potential applications.@xcite one example is the use of fano resonances to design plasmonic sensors.@xcite in this case , the shift of very sharp peaks or valleys in the extinction spectrum , consequence of collective effects of interacting nps , have been proposed to measure distances or the dielectric constant of the surrounding medium.@xcite more exotic applications of nps coupling are also possible and many of them are now being actively explored , such as clocking by metamaterials or plasmonics circuitry.@xcite although , linear arrays of metallic nps have been proposed as potential waveguides to transfer massive amounts of information at the nanoscale,@xcite nowadays it is clear that high damping factors would impose severe limits to that and most of the effort in this field is being put on overcoming this limitation . however , one important and very general feature of 1-d systems , which seems to have been overlooked in plasmonics , is the extreme sensitivity of their dynamical properties to slight changes of their parameters . in this respect ,
we have previously studied the plasmonics energy transfer from a locally excited np ( le - np ) , the 0th np in the scheme shown in [ figure-1 ] , to the interior of a semi - infinite np chain.@xcite this system presents a form of dynamical phase transition ( dpt ) , a phenomenon that is currently attracting interest as source of novel effects on various fields.@xcite basically a dpt in this context means that by sightly moving a single parameter , such as the shape of the nps or the distance between them , the system undergoes an abrupt transition from transferring all the injected excitation to the interior of the np array ( a delocalized state ) , to keeping all the external excitation in the nps closer to the excitation point ( a localized state).@xcite the aim of this work is to exploit this basic property into a realistic setting , in order to design a new kind of excitation - transfer plasmonic nanosensor ( etps ) .
the system studied , depicted in [ figure-1 ] , is modeled using the coupled dipole approximation .
@xcite in this model , each @xmath0-np is described by a dipole @xmath1 induced by the electric field produced by the other dipoles , @xmath2 , and the external source , @xmath3 .
we assume a generic ellipsoidal shape for the metallic nps and describe its polarizabilities in a quasi - static approximation,@xcite resulting in : @xmath4 } \left ( e_{i}^{(\mathrm{ext})}+\sum\limits_{j\neq i}e_{i , j } \right ) \label{pvector},\]]where @xmath5 is the volume , @xmath6 is a geometrical factor that depends on the shape of the np and the direction of @xmath7 , @xmath8 is the free space permittivity , and @xmath9 is the dielectric constant of the host medium around the @xmath10-th np . the dielectric constants @xmath11 of the nps are described by a drude - sommerfeld model : @xmath12 , where @xmath13 is the plasmon frequency , @xmath14 is the electronic damping factor , and @xmath15 is a material dependent constant that take into account the contribution of the bound electrons to the polarizability . the lsp oscillations over each np can only be transverse ( @xmath16 ) or longitudinal ( @xmath6 ) to the axis of the linear array and we assume , for simplicity , that the ellipsoids axes are aligned with respect to these @xmath16 and @xmath6 directions .
if the excitation wavelength is large compared with the separation between nps , @xmath2 can be taken in the near field approximation , @xmath17 , where @xmath18 , @xmath19 and @xmath20 is the distance between nps .
taking into account these considerations , @xmath1 and @xmath3 can be arranged as vectors @xmath21 and @xmath22 resulting in:@xcite @xmath23where @xmath24 is the dynamical matrix and @xmath25 is a diagonal matrix that rescales the external applied field according to local properties : @xmath26with @xmath27 } { \left[\epsilon _ { m , i}+l_{i}(\epsilon_{\infty}-\epsilon _ { m , i})\right]}$ ] . because of the cubic dependence of @xmath7 on @xmath20 , we neglect contributions beyond nearest neighbors,@xcite resulting in a tridiagonal @xmath24 . for nps belonging to the chain , @xmath28 @xmath29 and @xmath30 , and for the le - np , @xmath31 , @xmath32 and @xmath33 . the lsp frequencies @xmath34 and coupling constants @xmath35 , result : @xmath36 } , \label{omegasp}\ ] ] @xmath37 for spherical nps ( @xmath38 ) in vacuum ( @xmath39 ) within a drude - sommerfeld model which completely neglects contribution from bound electrons ( @xmath40 ) , eqs .
[ matrixp]-[omegax ] reduce to the simple form used in several references.@xcite this shows the meaning of @xmath41 , which essentially collects the deviations from the ideal model of spherical np in vacuum with @xmath42 , not only for @xmath43 but also for couplings between nps .
note that these deviations reduce to a constant for @xmath44 and/or small excitation frequencies @xmath41 ( strictly for @xmath45 \ll \omega_{_{\mathrm{p}i}}^{2}/ [ \epsilon_{\infty}-\epsilon_{m , i}]$ ] ) .
the form of eqs .
[ matrixp]-[omegax ] not only extends our previous expressions@xcite to non - spherical nps but also includes explicitly the interdependencies of @xmath35 , @xmath46 , and @xmath43 with the parameters of the system , _
i.e. _ size , shape , and material of nps , distances between nps , and dielectric constants .
it should also be mentioned that although this model can be further improved , the different corrections will only add quantitative corrections as long as the essential physics of the problem remains , _
i.e. _ a semi - infinite chain with nearest neighbors interactions.@xcite we will analyze this point in more detail in the last section . in the present work we will consider the case of excitation injection to a semi - infinte homogeneous linear array of nps , where solely the le - np can be different from the rest and
the only nonidentical separation is that between the le - np and the first np of the chain .
therefore , there will be two lsp resonances , that of the le - np @xmath47 and that of the chain s nps @xmath48 .
similarly , the couplings between the chain s nps will be all the same @xmath49 .
the couplings between the le - np and the first np of the chain are @xmath50 and @xmath51 .
they are not necessarily equal ( see eq .
[ omegax ] ) , since they describe quite different situations , _
i.e. _ how @xmath52 affects @xmath53 ( @xmath54 ) and how @xmath53 affects @xmath52 ( @xmath51 ) .
the values of the resonance frequency @xmath48 and the couplings @xmath55 can be tuned in different ways .
however , one must be careful because they are always interrelated .
for example , choosing a different material for the nps will change both @xmath56 and @xmath57 and therefore @xmath55 and @xmath48 . changing
the shape of the nps will modify @xmath6 and consequently @xmath55 and @xmath48 , except for @xmath58 in which case it will only modify @xmath48 .
the volume and distances between nps will change @xmath55 .
if the material , size and shape of nps are fixed , the free parameters that can be used for sensing purposes are the dielectric constant of the medium @xmath59 that alter both @xmath60 and @xmath55 , and the distances between nps that change couplings . the values of @xmath49 and @xmath48 define the passband which is the interval of frequencies at which an excitation can propagate with only a relatively small decay , given by @xmath14 , through the waveguide formed by the linear array of nps .
all excitations with frequencies @xmath61 outside the passband will decay along the chain exponentially and very fast . in the weak damping limit ( wdl )
@xmath62 , this passband is given by a simple expression , @xmath63 .
this fact shows one of the roles of the chain : to determine which frequencies can be propagated .
the other role is to perturb the local density of plasmonic states ( ldps ) of the le - np by pushing the resonance from its uncoupled value at @xmath47 to the nearest edge of the passband.@xcite this process occurs very abruptly when the resonance is about to cross the passband edge that is when the dynamics of the system changes completely.@xcite this phenomenon can be completely understood in terms of the divergences , or poles , of eq .
[ matrixp ] in the wdl.@xcite basically there are three different dynamical regimes of interest for excitation transfer : 1 ) _ delocalized _ or _ resonant _ state regime , where the maximum of ldps of the le - np falls within the passband and most of the excitation can be transferred to the np array ; 2 ) _ localized _ state regime where there is a very sharp peak in the ldps outside the passband and most of the excitation remains close to the excitation point ; and 3 ) a transition regime called _ virtual _ state where the ldps presents a non - lorentzian asymmetric peak just at the passband edge and most of the excitation can be transferred to the waveguide .
the transitions between the different dynamical regimes can be obtained analytically in the wdl and assuming frequency independent couplings , where:@xcite @xmath64 gives the _ resonant - virtual _ transition and : @xmath65 gives the _ virtual - localized _ transition , where @xmath66 , @xmath67 , and @xmath68 .
the constant @xmath69 is the relative coupling between le - np and first np of the chain in units of the coupling in the chain , @xmath70 yields the difference in resonance frequency of the le - np with respect to that of a chain s np , and @xmath71 measures the strength of the coupling between chain s nps in units of their own resonance frequency .
these formulas , although strictly valid only in the wdl , give an excellent estimation for finites @xmath14s.@xcite the excitation transferred from the le - np to the _
m_-th np of the chain can be calculated exactly ( in the quasi - static limit and within the nearest - neighbors approximation ) by using the formula:@xcite @xmath72 -\alpha \pi ( \omega ) \right ) } e^{-m ( l \pm iq ) } , \label{pm}\ ] ] where @xmath73 is the electric field on the le - np and the renormalized eigen - frequency @xmath74 is given by @xmath75 .
the decay length @xmath76 and the wavenumber @xmath77 depend on the self energy @xmath78 as : @xmath79 where @xmath78 is : @xmath80 - \label{pi } \\ &
\mathrm{sgn}(\omega ^{2}-\omega _ { _ { \mathrm{sp}}}^{2})\tfrac{1}{2}\sqrt{\left [ \omega ^{2}-\widetilde{\omega } _ { _ { \mathrm{sp}}}^{2}\right ] ^{2}-4\omega _ { _ { \mathrm{x}}}^{4}}. \notag\end{aligned}\ ] ] in all the plots of the next section with use @xmath81 as an indicator of the excitation transferred to the waveguide , with @xmath82 , @xmath83 , and using @xmath84 .
we have chosen @xmath84 , because around this value the behavior of @xmath81 vs @xmath85 becomes essentially a small decaying exponential for all @xmath61 within the passband or zero for excitation frequencies outside it .
thus , choosing a bigger value of @xmath85 would only add a multiplicative factor to the results , _
i.e. _ it would not change qualitatively the figures shown .
in this work , our purpose is to show how to use the sudden change in the dynamical properties of systems tuned around their dpt to device etpss . for this purpose
, we used the system depicted in [ figure-1 ] as an example .
[ figure-2 ] shows the maximum excitation transferred to a np of the waveguide , @xmath86 enabled by a variation of the excitation frequency @xmath61 , at each system configuration .
superposed are the critical values that separates dynamically distinct behaviors calculated in the wdl ( eqs .
[ alfac1 ] and [ alfac2 ] ) .
as it has been previously reported,@xcite excitation transfer is controlled by the dpt where _ virtual _ states are transformed into _
localized _ states ( gray dashed lines ) .
@xmath87{figure-3a.eps } \\ \includegraphics[width=2.5in]{figure-3b.eps}\end{array}$ ] analyzing the phase diagram of [ figure-2 ] it is clear that the strong dependence of excitation transfer to the system s parameters can be used for different forms of sensors . as a first application of this , lets us analyze the potentiality of this plasmonic device as a plasmon ruler . in [ figure-2 ]
a vertical displacement corresponds to a variation of the distance between the le - np and the chain .
the vertical blue arrow indicates an example of a region in the parameters space that can be used for such a purpose .
for each point along this arrow the system depicts a different excitation transfer spectrum .
three representative points along this line , at @xmath88 , @xmath89 , and @xmath90 , are used in [ figure-3]-*a * ) to illustrate how the spectrum of excitation transfer changes when the system undergoes a dpt .
note the abrupt change in the spectra induced by a slight change of @xmath91 : the excitation transfer suddenly increases at @xmath92 when @xmath91 varies from @xmath90 to @xmath89 , and then decreases equally abruptly when @xmath91 change from @xmath89 to @xmath93 .
this behavior is a consequence of the _ virtual - localized _ dpt , indicated in [ figure-2 ] by the curved gray dashed lines and predicted by eqs .
[ alfac1 ] and [ alfac2 ] .
one of the interesting aspects of this new form of plasmonic ruler is that the variation of a distance not merely modifies the position of a peak but dramatically changes the whole spectrum , and quite remarkably , before and after the dpt , only negligible signals should be observed independently of the excitation frequency . in this sense , our excitation transfer nanoruler could be used as a robust nanometric standard .
this is so , because , in order to observe a difference in the excitation transferred , it is not required a high accuracy in the excitation frequency nor a monochromatic light but keeping the maximum of the excitation frequency close to the lower band edge frequency , in our example at @xmath94 , is a sufficient condition .
furthermore , if the polarization has two components one transversal , @xmath16 , and one longitudinal @xmath95 , this should not interfere in a sensor tuned to work with only one of them .
this is so , because the passband of @xmath16 modes are half of @xmath6 modes , eq .
[ omegax ] , which imply that regions in the spectrum where dpts can be observed , should be well separated .
moreover , according to eqs .
[ alfac1 ] and [ alfac2 ] , dpts should occur for different values of system s parameters .
another important aspect is that the detection could be performed in a region encompassing several nps .
this is so , because , except for the first nps of the chain , the excitation transferred to the chain is an exponential function of np position @xmath85 , see eq .
thus , the main feature of this kind of nanorulers is its robust _ on _ and _ off _ feature , _
i.e. _ signals should only be detected when the le - np is at a very precise and predetermined distance .
@xmath87{figure-4a.eps } \\ \includegraphics[width=2.5in]{figure-4b.eps}\end{array}$ ] etps could also be used for sensing other properties , not only as plasmon nanorulers . as
shown in [ figure-2 ] , the system is also very sensitive to the value of the square frequency offset ( sfo = @xmath96 ) and consequently to @xmath97 , @xmath98 , and/or @xmath49 .
increasing the numerator of the sfo or decreasing its denominator , will both move the sfo away from the center of [ figure-2 ] as indicated by the horizontal green arrow .
[ figure-3]-*b * ) shows the excitation transfer spectrum for three values of sfo along this green arrow .
similarly to the nanoruler case , the spectrum experiences a significant change when moving the sfo along a dpt , from @xmath99 to @xmath100 . in order to show how to make profit of this feature we will consider two examples : one in which contraction or expansion of the whole system is measured , and one in which local dielectric constant around the le - np is measured .
[ figure-4]-*a * shows , for a fixed excitation frequency , @xmath94 , the relative increment of the excitation transferred , @xmath101 , as a function of the relative expansion / contraction of the whole system , which turns our device into an optical strain monitor@xcite or even into a temperature sensor , depending on the expansion coefficient of the host material .
note that a relative expansion / contraction of the whole system corresponds to a change in @xmath102 but not in @xmath103 , @xmath104 , nor in @xmath105 .
thus it is equivalent to an horizontal displacement in [ figure-2 ] .
[ figure-4]-*b * shows the variation of the relative increment of the excitation transferred as a function of the local dielectric constant around the le - np , a property potentially useful for molecular sensing purposes for example.@xcite in this case , a variation of @xmath106 changes not only @xmath107 but also @xmath108 and @xmath109 ( eqs . [ rii]-[omegax ] ) .
therefore , it is equivalent to a diagonal displacement in [ figure-2 ] and a rescaling of the external applied field .
the key ingredients from which the phenomenology described so far arise are : dominant nearest - neighbors interactions and the semi - infinite character of the system .
if this conditions are fulfilled , the results presented above should be applicable to the system under consideration besides quantitative corrections such as shifts of the spectra or the exact position of dpts in the parameter s phase space .
however , it is important to understand how our results are corrected by different factors . in this section
, we will study the effect of interactions beyond nearest - neighbors and retardation effects . in this case , eq .
[ pvector ] is still valid but all coupling between nps should be considered , _
i.e. _ @xmath24 is no longer tridiagonal , and these couplings should be determined by the dipole - induced electric field beyond the quasi - static limit , _
i.e. _ : @xmath110 \left ( 1- ikd \right ) \right\},\ ] ] where @xmath111 is the wavenumber in the dielectric , @xmath112 ( where @xmath113 is the speed of light in the medium ) , @xmath114 is the unit vector in the direction of @xmath115 ( where @xmath115 is the position of the observation point with respect to the position of the dipole ) . as the system consists of a linear array of nps where the spheroids axes are aligned with respect to the direction of the array , transversal ( @xmath16 ) and longitudinal ( @xmath6 ) excitations do not mix .
thus , the coupling terms @xmath116 of eq .
[ omegax ] acquire the form : @xmath117 where @xmath118e^ { ikd_{i , j } } \notag \\ & \widetilde{\gamma } ^{_{t}}_{i , j}=[1-ikd_{i , j}-(kd_{i , j})^ { 2 } ] e^ { ikd_{i , j } } .\end{aligned}\ ] ] eq . [ matrixp ] has now to be solved numerically which can be done by using standard methods of linear algebra .
@xcite the physical system considered in this section corresponds to a linear array of very flat oblate spheroidal nps of silver ( @xmath119 , and @xmath120x@xmath121rad / s ) with the radio of the minor axis equals to @xmath122 nm and the radii of the major axis equal to @xmath123 nm ( shape factor @xmath124 , ) .
the major to minor axis ratio is about 10 which is large enough to ensure a strong quadrupole quenching .
@xcite separation between nps of the chain is @xmath125 nm , from center to center , and they are aligned with the two equal axes perpendicular to the array direction . the external electric field is applied only to the first np where the direction of @xmath126 is also perpendicular to the array , _
i.e. _ @xmath127 .
electronic damping is calculated using the matthiessen s rule @xmath128 where the fermi velocity is @xmath129x@xmath130 m/s , the bulk mean free path is @xmath131 , we use @xmath132 , and the effective mean free path is calculated using @xmath133 where @xmath71 is the volume and @xmath134 the surface of the spheroids.@xcite for simplicity we consider a dielectric medium with @xmath135 . taking into account this , @xmath136 and @xmath137 give 0.032 and 0.41 ( for @xmath138 ) respectively .
we use here a slightly smaller coupling than that of the previous section for reasons that will become clear in the context of [ figure-5 ] . as before ,
the observation point is fixed at the 8th np , and we use the value of @xmath139 as an indicator of the excitation transferred to the chain . due to the large interparticle coupling , finite size effects result so important , especially near the lowest band edge , that it is necessary to take very long size of chains .
we use 200 nps , a value large enough to ensure negligible finite size effects . in arbitrary units , as function of @xmath140 for @xmath141 , @xmath142 and @xmath143 .
different curves show numerical results within the near field approximation for : nearest - neighbor interactions , fifth - nearest - neighbor interactions , and a fully converged calculation .
the system is at the dpt for the curve with nearest - neighbor interactions.,width=240 ] [ figure-5 ] shows the effect of including interaction beyond nearest - neighbors .
the first effect of this is essentially a red - shift and a broadening of the passband .
the second effect is that the dpt is also shifted from @xmath141 to @xmath144 as can be seen in [ figure-6]-a .
however , it is still present and indeed , as the damping term is proportional to the frequency , the peak is increased .
note that if @xmath82 were used , the peak would disappear from the spectrum as it would be located at @xmath145 , showing an important effect of higher - order contributions to interactions . in [ figure-6]-a and b retardation effects on the dpt are compared . here
, retardation causes nothing else but an increase of the intensity of the peak , in the region of the spectrum where the dpt is observable .
this is because , in this region of the spectrum , around @xmath146 , @xmath20 is much smaller than @xmath111 ( @xmath147 ) .
of course at higher frequencies , for example at @xmath148 , this is not the case ( @xmath149 ) and the behavior of the system is completely beyond the near field approximation .
@xmath87{figure-6a.eps } \\ \includegraphics[width=2.5in]{figure-6b.eps}\end{array}$ ] in arbitrary units , as function of its position @xmath85 .
retardation effects as well as all interactions between nps are considered . here
@xmath144 and the excitation frequency is @xmath150 , which correspond to the peak of [ figure-6]-b , width=240 ] finally , [ figure-7 ] shows the value of @xmath81 for different nps of the chain .
note that even after including all contribution to the interaction among nps and retardation effects , @xmath81 still decays exponentially as predicted by the theory , although the decay rate is different mainly because the passband edge is considerably shifted .
in this work we have studied the potentiality for sensing purposes of the excitation transferred from a le - np to the interior of a semi - infinite np chain , when the system is very close to its dpt condition .
while most plasmonic sensors use the shift of the lsp resonance peaks or the photoluminescence quenching of hybrid plasmonic - fluorophores systems , here we introduce a new working principle .
basically , the idea is to take advantage of the abrupt change in the plasmonic energy transferred when a control parameter is slightly changed around a dpt .
we have shown that this kind of sensor has the unique characteristic of having an _ on - off _ switching property and a high sensitivity , which opens new possibilities to design plasmonic devices such as plasmonic circuits activated only under certain environmental conditions .
we have also addressed the effects of different corrections to the approximations used and have shown that if the system s parameters are chosen properly , the dpt should be observable and therefore useful for the sensing purposes .
99 stewart , m. e. ; anderton , c. r. ; thompson , l. b. ; maria , j. ; gray , s. k. ; rogers , j. a. ; nuzzo , r. g. _ chem .
rev . _ * 2008 * , 108 , 494 - 521 .
mayer , k. m. ; hafner , j. h. _ chem .
rev . _ * 2011 * , 111 , 3828 - 57 .
halas , n. j. ; lal , s. ; chang , w .- s . ; link , s. ; nordlander , p. _ chem .
rev . _ * 2011 * , 111 , 3913 - 61 .
perassi , e. m. ; canali , l. r. ; coronado , e. a. _ j. phys .
c _ * 2009 * , 113 , 6315 - 6319 .
encina , e. r. ; perassi , e. m. ; coronado , e. a. _ j. phys .
* 2009 * , 113 , 4489 - 4497 .
coronado , e. a. ; encina , e. r. ; stefani , f. d. _ nanoscale _ * 2011 * , 3 , 4042 - 59 .
encina , e. r. ; coronado , e. a. _ j. phys .
c _ * 2010 * , 114 , 3918 - 3923 .
rong , g. ; wang , h. ; reinhard , b. m. _ nano letters _ * 2010 * , 10 , 230 - 8 .
jain , p. k. ; huang , w. ; el - sayed , m. a. _ nano letters _ * 2007 * , 7 , 2080 - 2088 .
jain , p. k. ; el - sayed , m. a. _ phys .
c _ * 2008 * , 112 , 4954 - 4960 .
tabor , c. ; murali , r. ; mahmoud , m. ; el - sayed , m. a , _ j. phys .
* 2009 * , 113 , 1946 - 53 .
chergui , m. ; melikyan , a. ; minassian , h. _ j. phys .
* 2009 * , 113 , 6463 - 6471 .
funston , a. m. ; novo , c. ; davis , t. j. ; mulvaney , p. _ nano letters _ 2009 , 9 , 1651 - 1658 .
huang , c .- ping ; yin , x .- gang ; kong , l .- bao ; zhu , y .- yuan _ j. phys .
c _ * 2010 * , 114 , 21123 - 21131 .
liu , c. ; li , b. q. _ j. phys .
* 2011 * , 115 , 5323 - 5333 .
ben , x. ; park , h. s. _ j. phys .
c _ * 2011 * , 115 , 15915 - 15926 .
haldar , k. k. ; sen , t. ; patra , a. _ j. phys .
* 2010 * , 114 , 4869 - 4874 .
ray , p. c. ; fortner , a. ; darbha , g. k. _ j. phys .
* 2006 * , 110 , 20745 - 8 .
sen , t. ; haldar , k. k. ; patra , a. j. _ phys .
c _ * 2008 * , 112 , 17945 - 17951 .
griffin , j. ; ray , p. c. _ j. phys .
b _ * 2008 * , 112 , 11198 - 201 .
seelig , j. ; leslie , k. ; renn , a. ; kuhn , s.;jacobsen , v.;van de corput , m. ; wyman , c. ; sandoghdar , v. _ nano letters _ * 2007 * , 7 , 685 - 689 .
singh , m. p. ; strouse , g. f. _ j. am .
soc . _ * 2010 * , 132 , 9383 - 91 .
anker , j. n. ; hall , w. p. ; lyandres , o. ; shah , n. c. ; zhao , j. ; duyne , r. p. van _ nature materials _ * 2008 * , 7 , 442 - 53 .
hodges , m. d. ; kelly , j. g. ; bentley , a. j. ; fogarty , s. ; patel , i. i. ; martin , f. l. ; fullwood , n. j. _ acs nano _ * 2011 * , 5 , 9535 - 41 .
ando , j. ; fujita , k. ; smith , n. i. ; kawata , s. _ nano letters _ * 2011 * , 11 , 5344 - 8 .
baca , a. j. ; montgomery , j. m. ; cambrea , l. r. ; moran , m. ; johnson , l. ; yacoub , j. ; truong , t. t. _ j. phys .
c _ * 2011 * , 115 , 7171 - 7178 .
maier , s. a. _ plasmonics : fundamentals and applications _ ; springer press : new york , 2007 .
novotny , l. ; hecht , b. _ principles of nano - optics _ ; cambridge press : cambridge , 2007 .
burin , l. ; cao , h. ; schatz , g.c . ; ratner , m.a .
b _ * 2004 * , 21 , 121 - 131 .
guillon , m. _ opt .
express _ * 2006 * , 14 , 3045 - 3055 .
zou , s. ; schatz , g. c. _ nanotechnology _ * 2006 * , 17 2813 - 2820 .
hernndez , j. v. ; noordam , l.d . ; bobicheaux , f. _ j. phys .
b _ * 2005 * , 109 , 15808 - 15811 . backes , t. d. ; citrin , d. s. _ phys . rev .
b _ * 2008 * , 78 , 153407 .
gozman , m. ; polishchuk , i. ; burin , a. _ phys .
* 2008 * , 372 , 5250 - 5253 .
maier , s.;kik , p. ; atwater , h. ; meltzer , s. ; harel , e. ; koel , b. ; requicha , a. _ nature mat . _
* 2003 * , 2 , 229 - 232 .
citrin , d. s. _ nano letters _ * 2004 * , 4 , 1561 - 1565 .
alu , a. ; engheta , n. _ phys .
* 2006 * , 74 , 205436 .
malyshev , a. v. ; malyshev , v. a. ; knoester , j. _ nano letters _ * 2008 * , 8 , 2369 - 2372 .
zou , s. ; schatz , g. c. _ j. chem .
phys . _ * 2004 * , 121 , 12606 - 12 .
abajo , f. j. _ rev .
phys . _ * 2007 * , 79 , 1267 - 1290 .
markel , v. ; sarychev , _
b _ * 2007 * , 75 , 085426
. park , s. y. ; stroud , d. _ phys .
b _ * 2004 * , 69 , 125418 .
gharghi , m. ; gladden , c. ; zentgraf , t. ; liu , y. ; yin , x. ; valentine , j. ; zhang , x. _ nano letters _ * 2011 * , 11 , 2825 - 8 .
guo , x. ; qiu , m. ; bao , j. ; wiley , b. j. ; yang , q. ; zhang , x. ; ma , y. ; yu , h. ; tong , l. _ nano letters _ * 2009 * , 9 , 4515 - 9 .
brongersma , m. l. ; hartman , j. w. ; atwater , h. a. _ phys .
b _ * 2000 * , 62 , r16356-r16359 .
bustos - marn , r. a. ; coronado , e. a. ; pastawski , h. m. _ phys .
b _ * 2010 * , 82 , 035434 .
kottos , t. _ nature physics _ * 2010 * , 6 , 166 - 167 .
bender , c. m. ; boettcher , s. _ phys .
lett . _ * 1998 * , 80 , 5243 - 5246 .
dente , a. d. ; bustos - marn , r. a. ; pastawski , h. m. _ phys .
* 2008 * , 78 , 062116 .
rotter , i. _ j. phys .
a : mathematical and theoretical _ * 2009 * , 42 , 153001 ; garmon , s. ; ribeiro , p. ; and mosseri , r. _ phys .
e _ * 2011 * , 83 , 23 .
jones , r. _ phys . rev . _ * 1945 * , 68 , 93 - 96 .
pastawski , h. m. ; medina , e. _ rev .
. de fis . _
* 2001 * , 47 s1 , 1 - 23 ; and .
references therein .
we used a complex implementation of lu decomposition method .
see : press , w. h. ; teukolsky , s. a. ; vetterling , w. t. ; flannery , b. p. _ numerical recipes in fortran 77 : the art of scientific computing _ ; cambridge university press : cambridge , 1998 ; vol .
kelly , k. l. ; coronado , e. a. ; zhao , l. l. ; schatz , g. c. _ the journal of physical chemistry b _ * 2003 * , 107 , 668 - 677 .
coronado , e. a. ; schatz , g. c. _ j. chem .
* 2003 * , 119 , 3926 - 3934 . | dynamical phase transitions ( dpts ) describe the abrupt change in the dynamical properties of open systems when a single control parameter is slightly modified .
recently we found that this phenomenon is also present in a simple model of a linear array of metallic nanoparticles ( nps ) in the form of a localized - delocalized dpt . in this work
we show how to take advantage of dpts in order to design a new kind of plasmonic sensor which should own some unique characteristics .
for example , if it were used as plasmon ruler it would not follow the so called universal plasmon ruler equation [ _ nano letters _ * 2007 * , 7 , 2080 - 2088 ] , exhibiting instead an _ on - off _ switching feature .
this basically means that a signal should only be observed when the control / measured parameter , _
i.e. _ a distance in the case of plasmon rulers , has a very precise and pre - determined value . here
, we demonstrate their feasibility and unique characteristics , showing that they combine high sensitivity with this _ on - off _ switching feature in terms of different distances and local dielectric constants .
this property has the potentiality to be used in the design of new plasmonic devices such as plasmonic circuits activated only under certain environmental conditions .
keywords : plasmonics , one dimensional arrays , open systems , localization , plasmon rulers . |
the quasinormal modes ( qnms ) of black holes @xcite , present a discrete set of complex frequencies corresponding to the solutions of the perturbation equations .
the qnms depend only on black hole parameters but not depend on the initial perturbation .
in general , the imaginary part of the complex frequencies are associated with the decay timescale of the perturbation while the real part represents the actual frequency of the oscillation .
the qnms are of the great importance in black hole physics , gravitational wave astrophysics , and might be observed in existing or advanced gravitational wave detectors such as ligo @xcite , and virgo @xcite . during last decades , a lot of authors had focused on the qnms of black holes in asymptotically flat spacetime @xcite , asymptotically ( anti-)de sitter ( henceforth ads or ds ) spacetimes @xcite , scalar perturbation @xcite , gravitational perturbation @xcite , dirac perturbation @xcite , and electromagnetic perturbation @xcite , in different backgrounds . some pioneer work
have extended the investigation of qnms to higher dimensional spacetime @xcite .
indeed , the qnms contain information on the parameters of the underlying black hole and hence have significance in identifying black holes in four and higher dimensions .
recently , the possiblity of production of tev - scale black holes @xcite at particle colliders such as the large hadron collider ( lhc ) have attracted much attention to the models of low scale gravity @xcite .
black holes in higher dimensional spacetime have attracted a lot of interest within different theories such as supersymmetric string theory and different extra dimensional scenarios namely arkani - hamed , dimopoulos and dvali ( add ) @xcite and randall - sundrum @xcite ( rs ) brane world models ( for a review see @xcite ) . in this way , based upon the fact that our universe may be embedded in higher dimensional spacetime , much interest has been paid to the rs model to resolve the gauge hierarchy problem . in the context of the rs brane world scenario ,
there has been an attempt to construct numerical black holes , being successful only in working out small localized black holes @xcite and some remarks were made in terms of the ads / cft correspondence as to why finding black hole solution localized on the rs brane is so difficult that large black holes will not be static @xcite .
+ in this paper , we discuss the quasinormal modes ( qnms ) of charged brane world black holes in the randall sundrum model considering both the electromagnetic and tidal charges effects as well as the thickness of the bulk .
there are several techniques to compute the qnms , such as the potential fit @xcite , leaver s continued fraction method @xcite , wentzel - kramers - brillouin ( wkb ) @xcite , etc . here
, we study the qnms of charged black holes embedded on the rs brane world considering quantum gravity effects and using the 3th order wkb approximation as the results do improve for higher orders of wkb @xcite . we discuss in detail how the qnms are influenced by the parameters of higher dimensional black hole on bulk spacetime .
+ this paper is organized as follows : in section ii we set up the basic notion and formalism of the charged black hole localized on the rs brane world and we obtain the effective potential , solving 5-dimensional klein - gordon wave equation . in section iii
we compute quasinormal modes of the charged black hole in brane world using 3th order wkb approximation . in the last section a summary of results , conclusions and discussion
are given .
in order to obtain the charged black hole solution localized on the randall - sundrum ( rs)@xcite brane world , we start with the action of the rs model in ( 4 + 1 ) dimensions as follows @xmath0 } - \int { { d^4}x\left [ { \sqrt { - { g _ + } } { \lambda _ + } + \sqrt { - { g _ - } } { \lambda _ - } } \right ] } } \ ] ] where @xmath1 is a cosmological constant while @xmath2 and @xmath3 are tensions of the branes at @xmath4 and @xmath5 respectively which are two singular points on the orbifold @xmath6 located at @xmath7 and @xmath8 . here , two 3-branes are placed at these points and we assume that orbifold possesses a periodicity in the extra coordinate @xmath9 , identifying @xmath10 with @xmath9 . in this case , one can define the metric on each brane as @xmath11 we assume that the induced metric on the brane take the general form @xmath12 where @xmath13 is the lapse function which describes embedding geometry of the hyper surface spanned by @xmath14 during the evolution in the bulk spacetime and @xmath15 is a metric of unit @xmath16-sphere . among possible solution satisfying the equation of motion @xcite
, we consider charged black hole solution in the rs model in which our universe is viewed as a domain wall in asymptotically ads space . in this case , using the hamiltonian constraint and solving hamiltonian equation along with metric ansatz ( 3 ) , one can obtain @xmath17 here , @xmath18 is the mass of black hole . @xmath19 and @xmath20 are tidal and electromagnetic charges respectively and we use @xmath21 to denote the position of the outermost horizon which @xmath22 .
we are interested in charges effects on the shape of perturbed horizon by scalar field , and how far it extends into the fifth dimension and this could be measured by the ratio of the physical size of the apparent horizon . in this case , the only apparent horizon that appears during the @xmath9 evolution is located at @xmath23 .
it was shown in the appendix of @xcite that at @xmath23 , @xmath24 and @xmath25 evolve synchronously , that is , @xmath26 and analytically continuing back to original spacetime yields @xmath27 . therefore , here for simplicity , we set @xmath28 .
then , we obtain @xmath29.\ ] ] indeed , a randall sundrum type brane model is built up which has two branes at @xmath30 and @xmath31 .
the solution ( 5 ) describes a charged black hole placed on the hypersurface at the fixed extra coordinates which in this brane background , the scalar field with mass @xmath32 , satisfies the klein - gordon wave equation as @xmath33 which one can write it explicitly as follow @xmath34\\ \begin{array}{*{20}{c } } { } & { + { e^{4\kappa \chi}}{\partial _ \chi } } \end{array}\left ( { { e^ { - 4\kappa \chi}}{\partial _ \chi}\phi } \right ) - { m^2}\phi = 0 \end{array}\ ] ] with @xmath35 here , the separation of variables is easily possible if we set @xmath36 which is discussed in @xcite . in this case , the modes along the extra dimension are quantized by stable standing waves , and then the eigenvalue is naturally discretized as @xmath37 where the quantum parameter @xmath38 is @xmath39 where @xmath40 and @xmath9 is the thickness of the bulk . in this way
, the reduced form of the equation ( 7 ) reads @xmath41 we substitute four dimensional wave function resolving the field into spherical harmonics as follows @xmath42 which is the usual separation of variables in terms of a radial field and a spherical harmonic @xmath43 . in the region the submanifold given by the patch @xmath44 which we will treat the dynamics of fields in the black hole exterior , one can define a `` tortoise coordinate '' @xmath45 as follows @xmath46 where @xmath22 and the black hole horizon located at @xmath21 . in this case , the radial function of ( 12 ) satisfy a schrdinger wave equation which for each multiple moment reads @xmath47 where @xmath48 is the effective potential given by @xmath49 here , the spectrum of spherical harmonics @xmath50 is given by @xmath51 here , the tortoise coordinate @xmath52 is defined on the interval @xmath53 in such a way that the special infinity @xmath54 correspond to @xmath55 , while the event horizon correspond to @xmath56 .
the effective potential ( 15 ) is positively defined in the region @xmath57 and has the form of the potential barrier which approaches constant values at both spatial infinity and event horizon , i.e. outside the event horizon @xmath58 . [ 1 ] the effective potential @xmath48 versus radial coordinate @xmath59 with @xmath60 for different values of @xmath19 in presence of the fixed @xmath61 .
here , we adopt @xmath62 , @xmath63 , @xmath64.,scaledwidth=40.0% ] [ 2 ] the effective potential @xmath48 versus radial coordinate @xmath59 with @xmath60 for different values of @xmath20 in presence of the fixed @xmath65 . here , we adopt @xmath62 , @xmath63 , @xmath64.,scaledwidth=40.0% ] [ 3 ] the logarithmic imaginary and logarithmic real part of the quasinormal frequencies in the absence of the electromagnetic charge , @xmath66 , for various values of @xmath19 .
we adopt @xmath62 , @xmath63 , @xmath67.,scaledwidth=50.0% ] [ 4 ] the imaginary part ( left ) and real part ( right ) of the @xmath68 with respect to @xmath20 in the absence of the tidal charge , @xmath69 with adopted @xmath62 , @xmath63 , @xmath67.,title="fig:",scaledwidth=40.0% ] the imaginary part ( left ) and real part ( right ) of the @xmath68 with respect to @xmath20 in the absence of the tidal charge , @xmath69 with adopted @xmath62 , @xmath63 , @xmath67.,title="fig:",scaledwidth=40.0% ] [ 5 ] the imaginary part ( left ) and real part ( right ) of the @xmath68 with respect to the @xmath19 for different electromagnetic charge , @xmath20 , with adopted @xmath62 , @xmath63 , @xmath67.,title="fig:",scaledwidth=40.0% ] the imaginary part ( left ) and real part ( right ) of the @xmath68 with respect to the @xmath19 for different electromagnetic charge , @xmath20 , with adopted @xmath62 , @xmath63 , @xmath67.,title="fig:",scaledwidth=40.0% ] in this case , the five dimensional effective potential , @xmath48 , contains the quantum number @xmath70 which is higher and thicker than the four dimensional case when @xmath71 . in this manner ,
the effective potential is related to the mass of the black hole , @xmath18 , electromagnetic charge , @xmath20 , and tidal charge , @xmath19 . here
, we consider three cases : i ) no electromagnetic charge , i.e. , @xmath66 which the induced metric has a regular horizon if @xmath72 , ii ) no tidal charge , i.e. , @xmath69 , which the induced metric on the domain wall is reissner - nordstrm with a correction term @xcite , and iii ) both charges non - zero , i.e. , @xmath73 , @xmath74 .
figure 1 represent the typical variation of the effective potential @xmath48 for different values of the tidal charges in presence of fixed value of the electromagnetic charge .
it is obvious that increasing value of both tidal and electromagnetic charges causes decrease of the peak of the potential barrier ( see figure 1 and 2 ) .
[ 6 ] the real part of the @xmath68 vs. the imaginary part of the @xmath68 for various values of the electromagnetic charge , @xmath20 , we adopt @xmath62 , @xmath63 , @xmath67.,scaledwidth=50.0% ] [ 7 ] the imaginary part ( left ) and real part ( right ) of the quasinormal frequencies with respect to the thickness of the bulk , @xmath9 , for @xmath75 and @xmath76 . here
, we adopt @xmath62 , @xmath63.,title="fig:",scaledwidth=40.0% ] the imaginary part ( left ) and real part ( right ) of the quasinormal frequencies with respect to the thickness of the bulk , @xmath9 , for @xmath75 and @xmath76 . here
, we adopt @xmath62 , @xmath63.,title="fig:",scaledwidth=40.0% ]
quasinormal modes are fundamental vibration modes around a black hole and they are obtained by solving equation ( 14 ) under the appropriate boundary condition . in order to consider boundary condition for an incoming wave at th event horizon , one can find that the effective potential @xmath77 for @xmath78 . in this case , an outgoing wave at infinity the effective potential vanishes , @xmath79 , as the black hole tend to the minkowski spacetime at infinity .
therefore , one can write the boundary condition for incoming and outgoing waves as @xmath80 in this manner , the general formalism to find the quasinormal modes of the black hole whose effective potential has the form of a potential barrier like that of formula ( 14 ) , one can use wkb approximation which has been proposed by schutz and will @xcite , extended by iyer and will @xcite to the third order beyond the eikonal approximation , and subsequently developed by many people @xcite .
the wkb corrections up to sixth order have been used in @xcite . in this case ,
when @xmath81_{\max } } \ll \left [ { { \omega ^2 } - v\left ( { \pm \infty } \right ) } \right]$ ] , wkb has very good accuracy , i.e. if solutions of equation @xmath82 are very close to each other as two turning points and the total energy , @xmath83 is expanded to the taylor series near the maximum of the effective potential near by the turning points . in order to calculate the quasinormal modes @xcite , using the method called the @xmath84th order wkb approximation , one can get the formula for quasinormal frequencies as @xcite @xmath85 where @xmath86 is the @xmath87th order derivative of the potential @xmath88 and the first and the second of @xmath89 are given by @xmath90 here , @xmath91 for @xmath92 is defined as follows @xmath93 in this case , the parameter @xmath94 is called tone number of quasinormal modes .
we focused on @xmath95 to obtain fundamental quasinormal modes as it is known that the approximation is good in case of @xmath96 .
we calculated the quasinormal frequencies for various values of the electromagnetic and tidal charges , @xmath20 and @xmath19 , using the third order wkb method .
note that we fixed @xmath63 related to the quantum parameter @xmath38 and considered a massless scalar field ( @xmath97 ) . in the absence of the electromagnetic charge , @xmath66 , in case of the @xmath98 the logarithmic imaginary part and real part of the @xmath68 decrease when @xmath19 increases as well as when @xmath99 ( figure 3 ) . here , we draw attention , in case of the absence of the electromagnetic and tidal charges , @xmath100 , it is not possible to calculate the quasinormal mode as @xmath101 . therefore , we excluded this point .
we calculated the quasinormal frequencies for various values of @xmath20 in the absence of the tidal charge , @xmath69 . in this case , the imaginary part and real part of the @xmath68 decrease when the electromagnetic charge , @xmath20 decreases ( figure 4 ) .
in figure 5 , we plotted quasinormal frequencies , both the real part and imaginary part , of the massless scalar field for the mass @xmath62 for various values of the electromagnetic charge .
we took the angular parameter @xmath102 . in this case .
we plotted a curve by changing @xmath19 from -15 to 15 . in this manner
, both of the imaginary part and real part of the @xmath68 of the different values of the electromagnetic charge seems to converge to a value as we decrease the tidal charge @xmath19 .
obviously , the real part of the @xmath68 with respect to the imaginary part , has less variation when one increases the value of the electromagnetic charge , @xmath20 ( see figure 6 ) .
we also calculated the quasinormal frequencies for different values of the thickness of the bulk in fixed value of the electromagnetic and tidal charges .
it is also interesting to observe that the imaginary part of the quasinormal frequency increases when the thickness of the bulk increase while the real part of the @xmath68 decreases ( figure 7 ) .
in this paper , we have investigated quasinormal modes of massless scalar field of the charged black holes embedded in the randall - sundrum brane world .
we found that the five dimensional effective potential @xmath103 depends on the value of @xmath59 , integration constant related to @xmath104 mass , thickness of the bulk , @xmath9 , and the electromagnetic and tidal charges , @xmath20 and @xmath19 respectively . in this case , we have computed the quasinormal frequencies spectrum of the charged black holes localized in the randall - sundrum brane world , using the third order of wkb approximation .
it is shown the quasinormal spectrum depends on the electromagnetic and tidal charges .
we found , in the absence of the electromagnetic charge , @xmath66 , the logarithmic imaginary part and real part of the @xmath68 decrease when @xmath19 increases ( figure 3 ) .
we also found the imaginary part and real part of the @xmath68 decrease when the electromagnetic charge , @xmath20 decreases ( figure 4 ) .
we also have computed the quasinormal modes for different values of the electromagnetic and tidal charges in fundamental state @xmath105 .
we compare the real part and imaginary part of the quasinormal frequencies and their variations respect to the electromagnetic and tidal charges ( figure 5 and 6 ) .
we observed , that the imaginary part of the quasinormal frequency increases when the thickness of the bulk increase while the real part of the @xmath68 decreases ( figure 7 ) .
the paper is supported by universiti kebangsaan malaysia ( grant no.frgs/1/2014/sg05/ukm/01/1 for n. abbasvandi ) .
shahidan radiman would like to to acknowledge receiving a grant dpp-2015 - 036 from ukm .
00 amelino - camelia , g. , int . j. mod
d , 11 ( 2000 ) 35 ; amelino - camelia , g. , nature , 418 34 ( 2002 ) ; amelino - camelia , g. , int . j. mod .
d , 11 1643 ( 2002 ) ; kowalski - glikman , j. , lect .
notes phys . , 669 131 ( 2005 ) . | we study the quasinormal modes of the massless scalar field of charged black holes embedded in the randal - sundrum brane world using the third order wkb approximation .
we consider the effects of the electromagnetic and tidal charges on quasinormal frequencies spectrum for charged black hole black holes as well as the effect of the thickness of the bulk .
quasinormal mode , black hole , randall - sundrum , brane world 04.50.-h , 04.50.gh , 04.62.+b , |
thanks to the outstanding angular resolution of the x - ray telescopes aboard the _ chandra x - ray observatory _ , we are able to study the structure of pulsar - wind nebulae ( pwne ) , formed by relativistic outflows from pulsar magnetospheres . particularly detailed _
chandra _ images have been obtained for pwne around the crab pulsar ( mori et al .
2002 ; hester et al .
2002 ) , vela pulsar ( helfand , gotthelf , & halpern 2001 ; pavlov et al .
2001a ) , and psr b150958 ( gaensler et al .
each of these images shows an approximately axially - symmetric pwn morphology , with an extended structure stretched along the symmetry axis .
this suggests that such structures , usually called _ jets _ , are common to at least young pulsars .
most likely , these jets are associated with collimated outflows of relativistic particles along the pulsar s rotation axes .
investigations of pulsar jets allow one to understand the origin and properties of pulsar winds and their interaction with the ambient medium .
moreover , since jets have been observed in many astrophysical objects ( e.g. , agns , galactic microquasars ) , studying pulsar jets may shed light on the mechanism of jet formation in these objects as well . due to its proximity ( @xmath17 pc , from the annual parallax of the vela pulsar
caraveo et al . 2001 ; dodson et al .
2003a ) , the vela pwn is particularly well suited for studying the pulsar outflows .
this pwn was discovered with _
( harnden et al . 1985 ) and studied extensively with _
rosat _ and _ asca _ ( markwardt & gelman 1998 , and references therein ) .
the _ einstein _ and _ rosat _ observations with the high resolution imagers ( hris ) have established the overall kidney - bean shape of the vela pwn , with a symmetry axis approximately co - aligned with the pulsar s proper motion .
the first observations of the vela pwn with _
( pavlov et al . 2000 ; helfand et al .
2001 ) have shown a detailed structure of the nebula , including inner and outer arcs , northwest and southeast jets , and bright knots in the southeast part of the pwn .
furthermore , pavlov et al . ( 2001a ) found that the nebula elements were variable in position , brightness , and , perhaps , spectrum .
based on two observations with the _ chandra _
s advanced ccd imaging spectrometer ( acis ) , pavlov et al .
( 2001a ) noticed a long , curved extension of the northwest jet in approximately the same direction , well beyond the apparent termination point of the bright `` inner jet '' at its intersection with the outer arc .
this dim `` outer jet '' ( dubbed `` filament '' in that paper ) had a brighter `` blob '' close to the jet s end , which showed particularly strong displacement between the two observations taken 7 months apart .
to understand the nature of the variability of the vela pwn , we carried out a series of eight monitoring observations with the _ chandra _ acis .
these observations have confirmed the dynamical structure of the pwn , with most dramatic changes occurring in the outer jet .
moreover , we were able to detect an `` outer counter - jet '' , a much dimmer extension of the southeast ( counter-)jet , which is clearly seen only in the summed image , together with highly asymmetric diffuse emission . in the present paper
, we concentrate mainly on the highly variable outer jet , which has been detected in ten acis observations and three observations with the high resolution camera ( hrc ) , deferring a detailed analysis of the other pwn elements to a future work .
we describe the observations , their analyses , and observational results in 2 , and discuss possible interpretations in 3 .
the summary of our results and concluding remarks are presented in 4 .
series of ten acis observations of the vela pulsar and its pwn were carried out from 2000 april 30 through 2002 august 6 .
the dates of the observations and the exposure times are given in table [ obs ] .
in all observations the target was imaged on the back - illuminated chip s3 , less affected by the radiation - induced changes in ccd charge transfer efficiency than the front - illuminated chips . to image the whole pwn on one chip ,
the pulsar was offset from the acis - s aim - point by @xmath18 along the chip row in each observation . in the first two observations only half of the chip
was read out ( 1/2 subarray mode ) , with a frame time of 1.5 s , to reduce the pile - up from the bright pulsar .
the rest of the observations were done in the full array mode , with a frame time of 3.24 s. the telescope focal plane temperature for all acis observations was @xmath19 c. we used the ciao software , v.2.2.1 ( caldb 2.10 ) for the data analysis , proceeding from level 1 event files .
the energy and grade of each event were corrected for charge transfer inefficiency ( cti ) using the tool developed by townsley et al .
( 2000 ) . to examine the changes in the pwn surface brightness
, we have scaled the number of counts per pixel for each of the ten acis observations as @xmath20 , where @xmath21 is the effective exposure time of @xmath22-th observation , and @xmath23 s is the exposure time of the reference observation ( obsid 2813 ) .
in addition , we restricted the energy range to 1.08.0 kev to reduce the background contribution at higher energies and get rid of the trailed images ( readout streaks ) produced by the bright pulsar , whose spectrum is softer than that of the pwn . to produce adaptively smoothed images
, we use the ciao tool csmooth , with the smoothing scale such that the signal - to - noise ratio is between 4 and 5 .
applying the ciao tool wavdetect to the individual images , we found 15 point - like sources on the s3 chip ; 9 of them , including the vela pulsar , we seen in at least 4 images .
the rms deviation of the _ chandra _ pipe - line coordinates of 8 of the 9 sources is @xmath24 , while for one off - axis source it is about @xmath25 .
this source is likely extended because wavdetect finds two sources in some of the images .
therefore , we exclude this source and conclude that the astrometry is better than @xmath26 .
we will use this value as the uncertainity of the coordinate measurements . to increase the total time span and the sampling , we also included three archival _ chandra _ hrc observations in our analysis ( see table 1 for the obsids and exposure times ) .
in all the three observations the vela pwn was imaged on the hrc - i ( imaging ) plate .
pipe - line processed level 2 event files from the archive were used for the analysis , without further reduction .
we scaled the binned images to the 2000 january 20 exposure time and adjusted the color scale to reveal the similar amount of detail that was seen in the acis images . to produce the smoothed images ,
each image was binned at a 2 pixel by 2 pixel scale and then smoothed similar to the acis images .
[ powertable ] llccc panel & obs i d & instrument & date of observation & exposure time ( s ) + 1 & 1518 & hrc - i & 2000 - 01 - 20 & 49,765 + 2 & 364 & hrc - i & 2000 - 02 - 21 & 48,050 + 3 & 128 & acis - s & 2000 - 04 - 30 & 10,577 + 4 & 1987 & acis - s & 2000 - 11 - 30 & 18,851 + 5 & 2813 & acis - s & 2001 - 11 - 25 & 17,933 + 6 & 2814 & acis - s & 2001 - 11 - 27 & 19,870 + 7 & 2815 & acis - s & 2001 - 12 - 04 & 26,960 + 8 & 2816 & acis - s & 2001 - 12 - 11 & 18,995 + 9 & 2817 & acis - s & 2001 - 12 - 29 & 18,938 + 10 & 1966 & hrc - i & 2002 - 01 - 13 & 49,464 + 11 & 2818 & acis - s & 2002 - 01 - 28 & 18,663 + 12 & 2819 & acis - s & 2002 - 04 - 03 & 19,920 + 13 & 2820 & acis - s & 2002 - 08 - 06 & 19,503 + + ' '' '' the large - scale x - ray structure of the vela pwn is shown in the summed acis image of observations 28132820 in figure 1 .
the brightest , kidney - shaped part of the pwn ( white and yellow in fig .
1 ) is approximately symmetric with respect to the direction of the pulsar s proper motion ( pa @xmath27 , in the local standard of rest see caraveo et al .
2001 , and references therein ) .
this deep image clearly reveals a long ( @xmath28 , as measured from the pulsar ) outer jet , approximately in the direction of the proper motion , which is somewhat curved and brightened at its end .
a much fainter outer counter - jet is seen , for the first time , in the opposite direction .
finally , we see highly asymmetric diffuse emission feature southwest of the jet / counter - jet line , which is obviously connected to the `` main body '' of the pwn .
choosing different brightness scale and color scheme , one can resolve a fine structure within the bright core of the pwn .
it is depicted in figure 2 which shows a zoomed individual image of 2002 january 28 .
the most apparent elements of this structure , with the pulsar [ 1 ] at the center , are the inner arc [ 2 ] , the outer arc [ 3 ] , the inner jet [ 4 ] , and the inner counter - jet [ 5 ] .
the inner jet is directed northwest from the pulsar ( pa @xmath27 ) in the direction of the pulsar s proper motion , and the counter - jet is directed toward the southeast ( pa @xmath29 pavlov et al .
2000 ) . the bright core ( red and yellow in fig .
2 ) is surrounded by a `` shell '' [ 6 ] of diffuse emission ( green in fig . 2 ) .
the outer jet [ 7 ] looks like an extension of the much brighter inner jet , well beyond the apparent termination point of the inner jet at its intersection with the outer arc .
it is clearly seen on this image , contrary to the much dimmer outer counter - jet .
comparison of the individual images shows that all the pwn elements are variable , especially the outer jet . throughout this paper
, we will concentrate our analysis upon the dynamic outer jet , though the inner jet and counter - jet will be considered for comparison .
the outer jet extends away @xmath30 pc ( @xmath31 lt - yr ) from the pulsar , where @xmath32 is the distance to the pulsar in units of 300 pc .
the characteristic width ( diameter ) of the outer jet is about @xmath33 cm . because the outer jet is generally co - aligned with the inner jet , it is natural to assume that the inner and outer jets are connected at the intersection with the outer arc , although the dim outer jet can hardly be seen within the shell .
figure 2 shows a typical view of the outer jet structure .
it is not as straight as the inner jet ; in this image it slightly bends southward from the inner jet / counter - jet direction , and back to the north near the end .
other individual images ( figs .
3 and 4 ) show different , sometimes more extreme , bending ( e.g. , panel 2 in fig . 3 and panel 13 in fig . 4 ) .
the average ( background - subtracted ) surface brightness of the outer jet in the ten acis images is about 0.05 counts arcsec@xmath34 ks@xmath7 , in the 18 kev band , with extremes of variablilty from @xmath35 to @xmath36 , in the same units ( errors here and below are 1 @xmath37 errors ) .
it is seen from figures 1 and 2 , that the outer jet brightness varies along its length .
choosing yet another brightness scale ( figs . 3 and 4 ) ,
we resolve compact regions of enhanced emission within the outer jet ( blobs , hereafter ) , with brightness up to @xmath38 counts arcsec@xmath34 ks@xmath7 , typically 23 times brighter than the remainder of the outer jet . in turn , the outer jet is approximately 3 times brighter than the background on the northeast side of the jet , with brightness @xmath39 counts arcsec@xmath34 ks@xmath7 .
the background on the southwest side of the outer jet is about 50% brighter than that on the northeast side .
this difference is quite clearly seen from figure 1 , which also shows that the increased background is connected to a region of enhanced diffuse emission on the southwest side of the bright pwn .
the average surface brightnesses of the inner jet and counter - jet are 1.5 and 1.8 counts arcsec@xmath34 ks@xmath7 , respectively , ranging from 1.1 to 1.9 for the inner jet and from 1.3 to 2.8 for the counter - jet , with typical errors of 0.2 counts arcsec@xmath34 ks@xmath7 .
the background - subtracted surface brightness of the outer counter - jet is @xmath40 counts arcsec@xmath34 ks@xmath7 , as measured from the summed image in figure 1 .
figures 3 and 4 give the time sequence of the 3 and hrc - i observations for the outer jet .
each of the panels corresponds to the appropriate data set from table 1 .
the variability of the outer jet is clearly seen , in both the hrc and acis images . over the thirteen observations , with different periods of time between each of them , we distinguish three different types of variability in the outer jet .
first , the outer jet shifts from side to side , bending and apparently twisting .
second , the blobs move outward along the outer jet .
finally , the blobs change in brightness and eventually disappear .
a very vivid example of the structural change in just one month is demonstrated by the first two _ chandra _ hrc observations ( panels 1 and 2 of fig .
3 ) . in panel 1
, the outer jet looks like a `` hook '' attached to the bright spot on the outer arc , with the outer end of the hook pointing due north and terminating at an elongated blob . in panel 2 ,
the hook evolved into a much more bent loop - like structure , which terminates with apparently the same ( albeit fainter ) blob at approximately the same location .
the structure has moved between two observations , indicating that the sideways shift ( e.g. , from point p in panel 1 to point q of panel 2 ) is approximately @xmath41 in the 32 days separating the observations .
this translates to a sideways speed of 0.6 c ( all the speeds in this section are given for @xmath42 pc ) .
another example is presented by the first two acis images ( panels 3 and 4 ) separated by 7 months .
the shape of the outer jet has changed from pointing due north and terminating with a blob to an s - shaped structure .
the displacement of the blob ( which easily could be another blob given that the time between the observations is so long ) suggests a speed @xmath43 . in observations 511 ( fig .
4 ) the jet appears almost straight , with a slight southward curve .
the short time ( 16 days ) between the fifth and eighth observations suggests that the sideways shifts of the jet occur on a time scale of order of weeks .
a dramatic change of the jet shape is again seen in the last observation 13 , separated by 4 months from the previous observation . in the last image ,
the jet points due west beyond the pwn shell and then abruptly turns to the north , with a bright blob at the turning point .
overall , the small - scale structural changes are seen in all of the observations .
for instance , the `` base '' of the outer jet ( where it leaves the shell see the white boxes in figs .
3 and 4 ) shifts from one observation to the next .
figure 5 shows smoothed one - dimensional distributions of counts across the outer jet ( i.e. , along the length of the narrow boxes in fig .
4 ) in eight consecutive acis observations .
typical apparent speeds of these shifts are of order a few tenths of speed of light .
the apparent speeds of the blobs along the jet are of similar magnitude .
the proper motions of the blobs a and b seen in several panels of figure 4 are shown in figures 6 and 7 .
the @xmath44 uncertainty of the position of the blobs is obtained by adding in quadrature the 07 uncertainty from the astrometry and an additional uncertainty , of approximately the same value , from measuring the centroid of the blob from the smoothed images .
blob b vanishes in panel 9 , so its apparent speed is calculated from four observations .
the apparent speeds of blobs a and b are @xmath45 and @xmath46 , respectively .
blob c vanished quickly , having an ( unconstrained ) apparent speed of @xmath47 .
assuming that the measurements of the apparent velocities of the blobs inside the outer jet give the bulk flow speed , @xmath48@xmath4 , and the jet does not deviate strongly from the sky plane , we obtain @xmath49@xmath50 yr for the travel time along the jet .
the actual travel time may be significantly different if the jet direction strongly deviates from the sky plane and relativistic effects are important ( see 3.1 ) .
the blobs change brightness as they move and disappear .
for instance , blob c slightly brightens in two days between observations 5 and 6 , then fades somewhat between observations 6 and 7 , and disappears completely after another week ( by observation 8) .
similarly , the brightness of blob b changes noticeably through observations 58 , while this blob vanishes in observation 9 .
no such blobs are found in either the inner jet or the counter - jet . for the spectral analysis of the jets in the acis data
, we used the xspec package , v.11.2.01 . in all spectral
fits we used the hydrogen column density fixed at @xmath51 cm@xmath34 , as determined from the observation of the vela pulsar with the _ chandra _ low - energy transmission grating spectrometer ( pavlov et al .
furthermore , because of a build up of a contaminant on either the acis ccds or the filters , additional time - dependent absorption is accounted for in the spectral fits using the acis_abs model ( see plucinsky et al .
the degree of absorption was set by the number of days since _
chandra _
launch . only photons in the energy range of 18 kev
were used for fitting to avoid contaminations from readout strips from the piled - up pulsar and to reduce the charged particle background . the photons from the outer jet
were extracted using polygons ( of somewhat different areas for the different observations see table 2 ) which enclosed the entire structures beyond the pwn shell .
two background regions were chosen the first from a dark region north of the outer jet ( background a ) , and the second sandwiching the outer jet ( background b ) .
various spectral models for emission from an optically thin thermal plasma were attempted to test if the emission associated with the outer jet is due to an interaction of the jet with the ambient ( snr ) gas .
the overall fits are formally acceptable , but the corresponding plasma temperatures , about 30 kev , are very high , in accordance with the lack of spectral lines in the spectra . at such high temperatures these models are equivalent to the thermal bremsstrahlung model , which , with the statistics available , is indistinguishable from a power - law at energies of interest .
the power - law model , @xmath53 , is more physical than the high - temperature bremsstrahlung because it naturally describes non - thermal ( e.g. , synchrotron ) radiation ; therefore , we use it for the spectral analysis of the jets .
an example of the power - law fit to a data set combined of observations 28132816 , chosen because of the similar extraction areas and locations on the chip , is shown in figure 8 , for background b. the results of the fits of the outer jet spectra with the power - law model for each of the individual acis observations are given in table 2 , for both backgrounds a and b. figure 9 shows the contours of the photon index @xmath54 versus spectral brightness at 1 kev , @xmath55 ( where @xmath56 is the area of the extraction region see table 2 ) . for each of the two backgrounds ,
the variability of the photon index is statistically insignificant , contrary to the variability of the brightness .
joint spectral fits ( fitting all data simultaneously with a common index and individual normalizations ) give @xmath57 and @xmath58 for backgrounds a and b , respectively , coinciding with the weighted average indices that are calculated from the individual fits for each of the two backgrounds . the difference between the average indices can be ascribed to systematic effects associated with the choice of background . since it is not immediately clear which of the two backgrounds is more adequate , we adopt a conservative estimate , @xmath59 , consistent with both background choices .
lcc & background a & background b + [ cols="<,^ " , ] + area is the polygon extraction area of the outer jet in arcsec@xmath60 .
normalization of the power - law spectrum , @xmath61 , is in units of @xmath62 photons cm@xmath34 s@xmath7 kev@xmath7 at 1 kev .
mean surface brightness , @xmath63 , is in units of 10@xmath34 counts arcsec@xmath34 ks@xmath7 , in the 18 kev range .
we searched for a change in power - law index along the length of the outer jet by extracting regions near the beginning of the outer jet , where it exits the shell of the bright pwn , and at its end . because of the low number of counts , a joint fit of all the acis observations was attempted .
the normalizations for each observation were allowed to vary ( because the extraction areas were different ) , while the photon index for all observations was fit to a common value .
the fits to these combined spectra were statistically acceptable , with @xmath64=1.19 ( 82 degrees of freedom [ dof ] ) and @xmath65 ( 96 dof ) for the beginning and the end of the outer jet , respectively ( using background b ) .
the change of the index , from @xmath66 at the jet beginning to @xmath67 near its end , is not statistically significant .
we also attempted to compare the spectra of the blobs with each other and with the spectrum of the rest of the outer jet .
for individual observations , no differences are seen because of the low number of counts .
combined fits , using background b , yield the photon indices @xmath68 ( @xmath64=1.01 , 70 dof ) for blob a ( observations 28132818 ) , @xmath69 ( @xmath64=0.47 , 20 dof ) for blob b ( observations 28132816 ) , and @xmath70 ( @xmath64=1.27 , 18 dof ) for blob c ( observations 28132815 ) .
thus , the spectral slopes do not show statistically significant differences even for the combined fits . for the inner jet and inner counter - jet
, we extracted the source counts from stretched elliptical areas of 12.7 and 18.5 arcsec@xmath60 , respectively , which enclose as much of the jet / counter - jet emission as possible without contamination from the other structures within the pwn , such as the arcs .
because of the very nonuniform distribution of surface brightness around the inner jet and counter - jet , we had to take backgrounds from small regions ; we chose circles ( 6.8 and 10.6 arcsec@xmath60 for the inner jet and inner counter - jet , respectively ) in the immediate vicinities northeast and southwest of the source extraction regions .
the average values of the photon index for the inner jet and inner counter - jet , @xmath71 and 1.20 , respectively , appear somewhat smaller than that for the outer jet .
however , the photon indices measured in individual observations show large deviations from the average values ( rms deviations are 0.24 and 0.10 , respectively ) , likely associated with statistical fluctuations of the bright background estimated from small regions .
therefore , we can not firmly establish the spectral differences between the outer jet and its inner counterparts from the data available .
as the outer counter - jet is hardly seen in the individual images , its spectrum was measured in the joint fit using observations 28132820 .
the source counts were extracted from a polygon area of 336 arcsec@xmath60 . because of the faintness of the outer counter - jet , the fit is rather sensitive to the choice of background . taking background from a region northeast of the end of the outer counter - jet , we obtain the photon index @xmath72 , while the index is smaller , @xmath73 , for the background from a region which sandwiches the outer counter - jet . thus , the spectrum of the outer counter - jet remains poorly constrained , its slope being indistinguishable from that of the outer jet , within the uncertainties .
assuming isotropic radiation , the average ( unabsorbed ) luminosities of the inner jet , inner counter - jet , and outer counter - jet , in the 18 kev band , are 2.6 , 4.5 and 0.5 @xmath74 erg s@xmath7 , respectively , for @xmath42 pc .
the average luminosities are computed from the average spectral parameters of the various components .
the outer jet demonstrates some variability in luminosity .
for background a , the extremes of the luminosity are @xmath75 and @xmath76 @xmath74 erg s@xmath7 , with an average of @xmath77 erg s@xmath7 . for background
b , the outer jet luminosity varies from @xmath78 to @xmath79 @xmath74 erg s@xmath7 , with an average of @xmath80 erg s@xmath7 . for comparison ,
the luminosity of the whole pwn within @xmath9 of the pulsar is @xmath81 erg s@xmath7 , in the same energy band .
the detected variability of the outer jet is the most vivid demonstration of the dynamical behavior of the vela pwn .
the observed changes of the overall shape of the outer jet over a time scale of weeks suggest apparent speeds up to ( 0.30.7)@xmath82 , which is comparable to the apparent speeds of the blobs moving along the outer jet away from the pulsar .
the spectra of the outer jet , as well as those of the other pwn elements , fit well with a power - law model , indicating that this is radiation from ultrarelativistic particles . these observations allow one to assume that the outer jet is associated with a polar outflow of relativistic particles from the pulsar s magnetosphere .
possible implications of our observational results are discussed below .
the orientation of the jets is the most important item for explaining the observed dynamical behavior , evaluating the velocities , and calculating the energetics of these outflows . immediately after the first _
observations of the vela pwn , it was suggested that the bright ( inner ) jet and counter - jet are co - aligned with the rotational axis of the pulsar and , presumably , with the direction of pulsar s velocity ( pavlov et al . 2000 ;
helfand et al .
2001 ) . as we have seen from figures 3 and 4
, the projection of the outer jet on the sky plane may strongly deviate from the ( straight ) extension of the projection of the inner jet / counter - jet .
however , the outer jet projection was close to that line during the period of 2001 november 25 through 2002 january 13 , when the apparent velocities of the bright blobs were measured . assuming , by analogy with the jets in agns and galactic microquasars , that the apparent motion of blobs is associated with motion of matter along an approximately straight line ( e.g. , the three - dimensional velocities of the blobs coincide with the bulk flow velocity ) , one can constrain the true speed , @xmath83 , and the angle @xmath84 between the line of sight and the direction of motion . the apparent speed , @xmath85 ,
is related to the true speed by the equation @xmath86 .
figure 10 shows the dependence @xmath87 for typical @xmath88 inferred from the observed motions of the blobs . for @xmath89 ,
the blobs can be approaching ( @xmath90 ) as well as receding ( @xmath91 ) .
the range of minimum values of @xmath92 is @xmath93 ( at @xmath94 ) , while the range of maximum allowed @xmath84 ( at which @xmath95 ) , is @xmath96 .
the latter values are smaller than the @xmath97 , inferred by helfand et al .
( 2001 ) under the assumption that the pwn arcs are dopler - brightened parts of a torus around the jet / counter - jet , and they are smaller then the @xmath98 suggested by pavlov ( 2000 ) assuming the arcs are doppler - brightened parts of ring - like shocks in relativistic conical outflows .
if we adopt , e.g. , @xmath99 , then the maximum possible apparent speed ( at @xmath95 ) is @xmath100 , which is below the lower limits of @xmath101 and @xmath102 on the apparent speeds of blobs a and b at @xmath42 pc .
thus , if any of the above - mentioned interpretaions of the arcs and the inner jet / counter - jet are correct , we have to conclude that the outer jet is tilted from the pulsar rotation axis towards the observer even if its projection onto the sky looks almost straight , like in panels 59 of figure 4 .
moreover , we can not rule out the possibility that the outer jet in these observations is strongly bent in the plane perpendicular to the sky plane e.g. , similar to the jet s sky projections in panels 1 , 2 or 13 . in this case
the directions of the local flow velocities vary along the jet , which might explain the different apparent velocities of the blobs a and b. such bending opens up a possibility that the bright regions of the jet are merely due to the projection effect the segments of a uniformly bright , bent jet , which are oriented along the line of sight , appear brighter because of the increased optical depth , and the brightness is further enhanced by the doppler boosting if the flow in these segments is streaming toward the observer .
for instance , if the loop - like jet in panel 2 of figure 3 were a two - dimensional stucture , the southern segment of the loop would look like a bright blob if it were observed from the southwest direction .
if this interpretation is correct , then the observed motions are caused by changing geometrical shape of the jet rather than the flow of locally bright material along the jet .
an argument against such an interpretation is that the outer jet apparently terminates with a bright blob in at least 10 of 13 observations , and it is hard to believe that the end segment of a randomly bent jet is so often directed toward the observer .
however , the extension of the jet beyond these bright blobs , as seen in the summed image of figure 1 ( see also 3.3 ) , suggests that this may still be the case . another interesting fact related to the jet geometry
is that the outer jet is consistently brighter than the outer counter - jet , with the ratio of surface brightnesses @xmath1039 ( the large uncertainty in the brightness ratio is due to the nonuniform backgrounds ) .
if the two jets are energetically similar and , on average , are oriented along a straight line , then the outer jet is approaching while the outer counter - jet is receding , contrary to the receding inner jet and approaching inner counter - jet in the geometry assumed by pavlov et al .
( 2000 ) and helfand et al .
( 2001 ) . within this interpretation ,
the brightness ratio is @xmath104^{\gamma + 2}$ ] , where @xmath54 is the photon index . for @xmath1039 and @xmath1051.4 , it gives @xmath1060.33 , @xmath107@xmath108 .
furthermore , these values of @xmath109 are consistent with @xmath1100.7 , inferred from the blob s motion , if @xmath111 and @xmath112 ( see fig .
10 ) . to reconcile this result with the previously suggested orientation of the inner jets , one has to assume that the outer jet is tilted by a large angle ( @xmath113@xmath114 ) from the inner jet direction towards the observer , while the outer counter - jet is tilted by a similar angle away from the observer .
contrary to the observed variable bendings of the outer jet , such tilts would have persisted for at least 5 years , since we confirm the existence of the brighter outer jet as far back as the 1997 _ rosat _ hri observation ( fig .
it is not clear what could cause such persistent tilts .
furthermore , it not clear why the outer jet and counter - jet are so much dimmer than their inner counterparts , and what causes the abrupt decrease of brightness in the outer jet and counter - jet , with bright knots at the junctions of the inner and outer parts .
in particular , the transition from the inner jet to the outer jet apparently occurs at the intersection of the jet with the outer arc , which hints that the outer arc is not a part of a ring - like structure but an umbrella - like shell pierced by the jet ( kargaltsev et al .
such a shell could be a bow shock created by the ( inner ) jet in the ambient medium , or it could form as a result of compression of the plasma outflow ( hence , amplification of the frozen - in magnetic field ) by the pulsar s motion in the ambient snr matter ( similar to that suggested by aschenbach & brinkmann 1975 for the crab pwn ) .
another possible effect of the pulsar s motion is distortion of the ring - like structure(s ) ( e.g. , the post - shock region presumably associated with the inner arc ) by the ram pressure of the ambient matter , if the pulsar s rotational axis is not co - aligned with its velocity . in this case the leading part of the ring can be flattened by the ram pressure so that the assumption that the ring s sky projection is a perfect ellipse can result in wrong values for the inclination angles .
finally , one could consider a possibility that the structures appearing as the bright inner jet and counter - jet are , in fact , merely traces of particle beams in conical outflows brightened by the doppler boosting ( e.g. , radhakrishnan & deshpande 2001 ; pavlov 2000 ) , while the outer jet and counter - jet are the `` true pulsar jets '' .
such an assumption implies a quite different interpretation of the whole pwn . we will discuss these and other possible pwn models in a future paper
currently , we can not unambiguously determine the orientation of the jets or the true velocities .
however , from the observed variations of the shape of the outer jet , with mildly relativistic apparent velocities , we can conclude that the true velocities are neither ultrarelativistic nor nonrelativistic ; @xmath115 and @xmath116 can be adopted as conservative lower and upper limits .
the power - law spectra of the outer jet and the other pwn elements can be interpreted as optically thin synchrotron emission from ultrarelativistic electrons ( and/or positrons ) with a power - law spectrum : @xmath117 , where @xmath118 is the lorentz factor , @xmath119 .
the electron index @xmath120 is related to the photon index as @xmath121 , i.e. , @xmath1221.8 for @xmath1231.4 observed in the outer jet .
a characteristic energy @xmath124 of the synchrotron photon depends on @xmath125 and magnetic field @xmath126 as @xmath127 kev , where @xmath128 , @xmath129 .
if the minimum and maximum energies of the photon power - law spectrum are @xmath130 and @xmath131 , respectively , then the corresponding boundaries of the electron power - law spectrum are @xmath132^{1/2 } b_{-4}^{-1/2}$ ] and @xmath133^{1/2 } b_{-4}^{-1/2}$ ] , where the photon energies are in units of kev , @xmath134 is the doppler factor ( @xmath135 is the speed of the bulk motion ) , and @xmath136 and @xmath137 are dimensionless factors .
the values of these factors are given in table ii of ginzburg & syrovatskii ( 1965 ; gs65 hereafter ) ; e.g. , @xmath138 and 1.8 , @xmath139 and 0.032 , for @xmath140 and 2.0 , respectively .
we will use the well - known formulae for synchrotron radiation to estimate the magnetic field and energetics of the outer jet .
since the bulk motions observed in the jet are only mildly relativistic ( @xmath141 is a plausible estimate ) , the doppler factor is not strongly different from unity ; moreover , its actual value is unknown because of the uncertain orientation of the jet .
therefore , we will neglect the bulk motion of the jet s matter for most of the estimates below . from the _ chandra _ acis observations ,
the power - law spectrum of the outer jet is seen in the energy range between @xmath142 kev and @xmath143 kev . since @xmath144 and @xmath145 , we can constrain the minimum and maximum energies of the electron power - law distribution : @xmath146 , @xmath147 .
one can put a lower limit on the magnetic field in the outer jet using the condition that the larmor radius of most energetic electrons responsible for the x - ray emission , @xmath148 cm , does not exceed the jet radius , @xmath149 cm .
this condition gives @xmath150 @xmath11 g .
an upper limit on the magnetic field can be estimated from the fact that the spectrum of the outer jet maintains its power - law shape ( shows no appreciable spectral break ) at the observed energies @xmath151 kev up to the end of the outer jet .
this means that the time of synchrotron losses , @xmath152 s , for the most energetic particles responsible for the observed x - ray spectrum [ @xmath153 is longer ( or comparable to ) the flow time , @xmath154 s , where @xmath155 cm .
this gives @xmath156 @xmath11 g , i.e. , @xmath157300 @xmath11 g for plausible values of @xmath158 and @xmath92 .
an independent estimate on the plasma parameters for the outer jet can be obtained from the brightness of the synchrotron radiation , assuming some value for the ratio @xmath159 of the magneic energy density , @xmath160 , to the energy density of relativistic particles , @xmath161 .
we can not exclude the possibility that there are some relativistic ions in the jet , in addition to the synchrotron - emitting electrons ( e.g. , gallant & arons 1994 ) , so that @xmath162 , where @xmath163 is the ratio of the energy density of ions to the energy density of electrons , @xmath164 .
making use of the standard formulae for synchrotron radiation ( e.g. , gs65 ; pacholczyk 1970 ) , one can express the magnetic field in terms of the observable parameters . for a tangled magnetic field ( randomly distributed along the line of sight )
, we obtain @xmath165 \frac{\mathcal{b}_{-7}}{\bar{s}_{16}}\right\}^{2/7}\,\,\mu{\rm g}\,,\ ] ] where @xmath166 photons ( scm@xmath60kevarcsec@xmath60)@xmath7 is the average spectral surface brightness at @xmath168 kev , @xmath61 is the normalization of the photon spectral flux measured in area @xmath56 , @xmath169 cm is an average length of the radiating region along the line of sight , @xmath170 , @xmath171 , @xmath130 and @xmath131 are the lower and upper energies of the photon power - law spectrum ( in kev ) , and @xmath172 is a numerical coefficient given by eq .
( 3.32 ) and table ii of gs65 ( e.g. , @xmath173 and 0.117 for @xmath122 and 1.8 [ @xmath105 and 1.4 ] , respectively ) .
the value of @xmath126 depends on several unknown parameters .
particularly uncertain are the boundary energies of the photon power - law spectrum because of the lack of high - resolution observations outside the x - ray range .
fortunately , the dependence on these energies is rather weak : @xmath174 for @xmath175 ( which implies @xmath176 ) , and @xmath177^{2/7}$ ] , for @xmath178 . for @xmath179 ( a typical spectral surface brightness at 1 kev
see fig .
9 ) , @xmath18010 ( the main reason of the uncertainty is the unknown spatial structure and orientation of the jet ) , and @xmath1811.4 , we obtain @xmath182^{-2/7 } \la 200\,\mu{\rm g}$ ] , for plausible ranges of @xmath131 and @xmath130 .
it is usually assumed that @xmath183 ( equipartition condition ) , while the value of @xmath163 is rather uncertain .
for the lower limit on the equipartition field not to exceed the upper limit estimated from ( the lack of ) synchrotron cooling , @xmath163 should not exceed @xmath184 .
thus , given the uncertainty of the parameters , we can only state that the equipartion field is consistent with the above - estimated limits , and that a plausible estimate for a typical field in the jet is @xmath185 g , with an uncertainty of a factor of 3 .
it should be noted that the local field values can differ from the average field .
for instance , if the brightness of the blobs is due to an increased magnetic field , at fixed @xmath186 , then the field in the blobs is a factor of 1.4 higher than the average field in the jet . on the other hand ,
the higher brightness can be explained by the projection effect ( larger @xmath186 ) and the doppler boost ( @xmath187 in eq . [ 1 ] if the bulk motion is taken into account ) .
assuming equipartition , @xmath188 erg @xmath189 , we can estimate the total energy of relativistic particles , @xmath190 erg , where @xmath191 @xmath192 is the volume of the jet . for plausible values of the magnetic field and the volume ( @xmath1931 , depending on the shape and orientation of the jet )
, we obtain @xmath194@xmath195 erg .
the energy injection rate and the energy flux can be estimated as @xmath196 erg s@xmath7 and @xmath197 erg cm@xmath34 s@xmath7 , respectively , where @xmath198 is the jet radius , and @xmath199 is the flow speed .
this injection rate is a small fraction of the spin - down energy loss rate of the vela pulsar , @xmath200 . on the other hand , it greatly exceeds the observed x - ray luminosity of the jet , @xmath201 , which means that most of the injected energy is emitted outside the x - ray range or , more probably , carried out of the jet without substantial radiation losses .
one can also estimate the electron ( positron ) injection rate , @xmath202^{-1}$ ] , where @xmath203 [ ( 2-p)[(\gamma_m^{1-p } - \gamma_m^{1-p})]^{-1}$ ] is the average electron lorentz factor for the power - law distribution . for the mean observed @xmath204 , we have @xmath205 if @xmath206 .
unfortunately , the minimum and maximum electron energies can not be determined without multiwavelength observations of the jet , therefore we will scale @xmath207 to a possible ( but arbitrary ) value of @xmath208 , which gives @xmath209 s@xmath7 .
this estimate corresponds to the mean electron number density @xmath210 @xmath189 .
the numerical estimates for @xmath211 and @xmath212 strongly depend on the boundary energies of the electron power - law spectrum ; in particular , the estimates become 3 orders of magnitude larger if @xmath213 is low enough ( @xmath184 ) for the jet to be synchrotron - emitting in the radio band .
it is interesting to compare the estimate for @xmath211 with the pair production rate expected for the vela pulsar : @xmath214 s@xmath7 , where @xmath215 is the goldreich - julian density , @xmath216 cm and @xmath217 s are the neutron star radius and spin period , @xmath218 g is the pulsar magnetic field , and @xmath219 is the pair multiplication coefficient ( its plausible value is @xmath220 see , e.g. , hibschman & arons 2001 ) .
the ratio @xmath221 can be interpreted as the fraction of pairs escaping from the pulsar through the outer jet , assuming that no pairs are created in the jet itself .
it is worth noting that the maximum lorentz factor for the pairs created in the pulsar magnetosphere does not exceed a few @xmath222 , for both the polar cap models ( harding , muslimov , & zhang 2002 , and references therein ) and the outer gap models ( e.g. , zhang & cheng 1997 ) .
the maximum lorentz factor in the outer jet , @xmath223 , is substantially higher , which means that the pairs , created by the pulsar , have been additionally accelerated beyond the pulsar magnetosphere .
if ions are pulled out of the neutron star surface layers , their outflow rate is constrained to the goldreich - julian value : @xmath224 s@xmath7 , while their characteristic energy may exceed that of electrons / positrons ( gallant & arons 1994 ) .
although the ions can be further accelerated to even higher energies , the maximum ion energy in the jet can not exceed @xmath225 ev [ @xmath226 because the larmor radius of ions with higher energies is larger than the jet radius .
assuming that the ion - to - electron number density ratio in the jet does not exceed that in the magnetosphere outflow , we can constrain the ion - to - electron energy density ratio : @xmath227 , at @xmath228 cm .
this means that relativistic ions do not make a substantial contribution to the energetics of the outer jet .
since the bulk velocities in the outer jet are only mildly relativistic , the bulk kinetic energy of relativistic particles is much lower than the energy of their random motion .
however , we can not rule out the possibility that some nonrelativistic electron - ion plasma from the ambient snr medium is entrained into the jet via interaction with the jet s magnetic field . to estimate an upper limit on the maximum density of the nonrelativistic component and maximum kinetic energy of the outer jet
, it seems reasonable to assume that the _ total _ energy injection rate , including the kinetic energy , into the outer jet can not exceed a fraction of @xmath229 of the total power , @xmath230 erg s@xmath7 , supplied by the pulsar , which implies @xmath231 erg .
this condition requires densities of nonrelativistic particles ( ions or electrons ) , @xmath232 @xmath189 .
thermal x - ray radiation from such a low - density plasma is orders of magnitude fainter than the observed radiation from the outer jet .
the corresponding upper limits on the pressure and energy density of the nonrelativistc components depend on the unknown mean energy of nonrelativistic particles ; however , they do not exceed the pressure and energy density of relativistic particles and the magnetic field .
furthermore , the density of this plasma is much lower than the density of the ambient medium . on the other hand ,
the bulk pressure , @xmath233 erg @xmath189 , can be higher than the pressure in the ambient medium , @xmath234 erg @xmath189 .
we see from figures 3 and 4 that the outer jet is never straight , showing either gentle bends ( e.g. , panels 510 ) or a strongly curved structure ( panels 1 , 2 , and 13 ) .
the varying curvature of the outer jet can be explained by its interaction with the ambient medium ( snr plasma ) and/or by a kink instability in the jet flow . a strong argument for an external wind to cause jet s bending is provided by the deep image of the pwn ( fig .
1 ) that allows one to see an extension of the outer jet beyond the apparent termination points ( often associated with outermost blobs ) observed in the individual images .
it shows that the outer jet does _ not _ terminate abruptly , but it rather smoothly bends clockwise ( towards west and then southwest ) by at least @xmath235 . such bending suggests a persistent _ northeast wind _ in the ambient medium ( approximately perpendicular to the direction of the pulsar s proper motion .
such a wind could also explain the very asymmetric diffuse emission outside the bright pwn the wind `` blows off '' relativistic electrons , produced in the bright pwn , towards the southwest .
moreover , it could explain , in the same way , the fact that the surface brightness is substantially higher at the southwest side of the outer jet compared with the northeast some high - energy particles are leaking from the jet and blown away by the wind . finally , an additional support for the wind comes from the radio image of the ( outer ) pwn ( lewis et al . 2002 ; dodson et al .
this image shows two lobes , northeast and southwest of the x - ray bright pwn , with a much smaller northeast lobe confined by a brightened northeast boundary .
the smaller size and the brightening can be explained by compression of the radio - emitting plasma by the external northeast wind .
the bulk pressure of this wind can be crudely estimated as @xmath236 , where @xmath237 is the diameter of the jet , @xmath238 is the curvature radius of the bending , and @xmath239 is the thrust ( jet s momentum flux ) through the jet s transverse area @xmath240 ( e.g. , leahy 1991 ) . if the bulk kinetic pressure is much lower than the magnetic pressure within the jet , @xmath241 erg @xmath189 , then the main contribution to the thrust comes from the longitudinal component of the magnetic stress tensor that , in turn , depends on the geometry of the magnetic field . for instance
, we obtain @xmath242 for a tangled magnetic field , which gives @xmath243 erg @xmath189 , for @xmath244 cm .
this bulk pressure is much lower than the typical thermal pressure in the vela snr .
it corresponds to the wind velocity @xmath245 km s@xmath7 , where @xmath246 is the ion ( proton ) number density in the wind .
attempts at separating the wind s x - ray emission from the snr background did not yield conclusive results . assuming a low - density wind , @xmath247 @xmath189
, we obtain @xmath248 km s@xmath7 high but not improbable velocity . such a velocity is comparable with the pulsar s proper motion velocity , @xmath249 km s@xmath7 ( caraveo et al .
2001 ) , but the pulsar s motion itself can not initiate the observed bending because the pulsar moves in the direction of the unbent jet ( at least in the sky projection ) .
however , once the jet is substantially bent from its original direction by the putative snr wind , the pulsar s motion with respect to the snr matter can bend it further , so that the jet can become directed backwards ( towards southeast ) .
a hint of such a bend is indeed seen in figure 1 .
moreover , the relativistic particles blown off the jet by the snr wind are being picked up by the head wind ( in the pulsar s reference frame ) and dragged in the direction opposite to the pulsar s proper motion , feeding the diffuse nebula .
such a picture is supported by the spectral slope of the diffuse emission , @xmath250 , which is softer than the emission of the brighter outer jet ( @xmath251 ) , but harder than the emission of the pwn shell ( @xmath252 ) .
thus , we can conclude that the outermost observable part of outer jet is likely bent by the combined action of the snr wind and the pulsar s proper motion , and the outer jet finally bends backward and get diffused southwest of the pulsar .
so far we have discussed only the bending of the jet s `` end '' .
explaining more extreme bends in the brighter part of the outer jet ( e.g. , the hook - like or loop - like structures in panels 1 , 2 , 4 , and 13 ) by the wind action is more problematic .
although we see that these bends are convex in the direction of the snr wind suggested above , the relatively small size ( and strong curvature ) of these structures would require a strong nonuniformity and very high velocities of the wind .
even a stronger argument is the observed variability of the bent structures , associated with almost relativistic velocities .
therefore , we have to invoke another mechanism to explain these features , not related to external winds .
a natural explanation is the kink instability of a magnetically confined , pinched flow .
estimates of the instability growth times and wavelengths depend on the model of such a flow , particularly the distribution of currents and topology of the magnetic field .
the flow model is also tightly connected with the interpretation of the whole pwn , particularly the inner jets and arcs .
analysis of these models is beyond the scope of the present paper .
therefore , we only briefly mention that the ( outer ) jet can be modeled as a plasma beam carrying a charge current of @xmath253 @xmath254 s@xmath7 ( @xmath255 see 3.2 ) , self - confined by a predominantly toroidal magnetic field ( z pinch ) . a similar model for pulsar jets
was suggested by benford ( 1984 ) , while collimation and confinment of agn jets by magnetic field was reviewed by begelman , blandford & rees ( 1984 ) and other authors .
growth times of mhd instabilities in such a flow are proportional to the alfven crossing time , @xmath256 ( e.g. , begelman 1998 ) . using the expression for the alfven velocity in the ultrarelativistic plasma , @xmath257^{-1/2}$ ] ( akhiezer et al . 1975 )
, we obtain @xmath258 at equipartition between the magnetic and particle energy densities , which gives @xmath259 days .
this time is comparable to the time scales of strong bendings ( @xmath260 days ) , which can be associated with the kink instabilities , and time scales of blob brightening ( about a week ) , which can be associated with the sausage ( neck ) instabilities .
the multiple _ chandra _ observations of the vela pwn have allowed us to investigate the dynamical outer jet and discover a dimmer outer counter - jet outside the bright pwn .
our main results can be summarized as follows .
the outer jet extends up to about 0.40.5 light years from the pulsar in the sky plane along the direction of the pulsar s proper motion .
its shape and brightness are variable on time scales of days to weeks .
the brightness is nonuniform along the jet , with brighter blobs moving away from the pulsar with apparent subrelativistic speeds .
the variations observed suggest typical flow velocities of 0.30.7 of the speed of light .
the outer jet is , on average , a factor of 7 brighter than the outer counter - jet .
if the outer jet and outer counter - jet are intrinsically similar but streaming in opposite directions , then the difference in brightness means that the outer jet is approaching at an angle of @xmath261@xmath262 to the line - of sight while the outer counter - jet is receding .
such an orientation apparently contradicts to the previously suggested models of the inner jets and the bright arcs .
the synchrotron interpretation of the hard ( @xmath263 ) power - law spectrum of the outer jet requires highly relativistic electrons or positrons , with energies of up to @xmath264 tev , and a typical magnetic field of about 100 @xmath11 g .
the outer jet s spectrum is perhaps slightly softer than those of the inner jet and counter - jet , but it does not change appreciably along the outer jet .
if the outer jet is a mildly relativistic outflow of an ultrarelativistic electron / positron plasma , the energy injection rate is @xmath265 . \4 . outside the bright pwn , there is an asymmetric , dim outer diffuse nebula that is substantially brighter southwest of the jet / counter - jet line .
its spectrum is softer than that of the outer jet , but it is harder than the spectrum of the brighter pwn shell .
it is possible that the x - ray emitting particles in the dim nebula are supplied through the outer jet , whose end part turns southwest with respect to its average ( northwest ) direction in the sky plane .
such a turn can be caused by a northeast snr wind with a speed of a few times 10 km s@xmath7 , which also helps feed the dim nebula .
the width of the outer jet , @xmath266 cm , remains approximately the same along the jet in different observations , including those which show strong bends .
this suggests an efficient confinement mechanism , perhaps associated with magnetic fields generated by electric currents in the pinched jet .
the current required , @xmath267 , is an order of magnitude lower than the goldreich - julian current in the pulsar magnetosphere .
the bright blobs and strong bends could be caused by the sausage and kink instabilities , respectively , in such a pinched jet .
the excellent resolution of the _ chandra _ telescope and high sensitivity of its detectors allowed us to obtain the spectacular pictures of the vela pwn , including the faint outer jet , and , in particular , to prove the highly anisotropic and dynamical nature of the pulsar outflows .
however , as has been often the case with new high - quality data , our results raise new questions and put under doubt the previous simplistic interpretations of pwne in general and the vela pwn in particular .
the most unclear issue is the interpretation of the complicated morphology of the vela pwn , particularly the relationship of the outer jet and outer counter - jet with their inner counterparts , the true orientation of the jets , and the actual topology of its arcs .
we do not even know the actual three - dimentional orientations of the inner and outer jets and counter - jets , nor do we understand the cause of their different brightness
. we can only guess about the nature of the bright blobs moving along the outer jet , the mechanism(s ) of the jet confinement , and the origin of the jet bendings .
at least some of these issues can be clarified by a series of deeper _ chandra _ observations of the vela pwn , taken with intervals of a few days , the now - established time scale of the pwn variations . in particular
, such observations would allow one to search for the spectral changes along the outer jet and check if the spectra of the blobs are different from those of the rest of the jet .
they could also help find blobs in the very dim outer counter - jet and measure their velocities , which would provide a clue to the geometry of the system .
the interpretation of the vela pwn would be much easier if it were detected at other wavelengths , outside the x - ray range .
so far , contrary to the much better studied crab pwn , the vela pwn has not been detected in the optical , mainly because of numerous relatively bright field stars and snr filaments in the pulsar vicinity ( mignani et al .
2003 ) . to get rid of their light and detect the pwn or put a stringent upper limit on its brightness , the field must be observed in polarized light .
it would also be very important to obtain a deep radio image the field around the vela pulsar with arcsecond resolution ( cf .
dodson et al .
both the optical and radio observations would provide estimates on the lower frequencies of the synchrotron spectra ( hence , lower energies of relativistic electrons ) , the critical parameters for evaluating the magnetic fields and the energetics of the observed pwn elements . measuring polarizations of the optical and radio emission would be crucial to establish the directions of the magnetic field in the pwn elements and understand their nature .
we thank leisa townsley and george chartas for the advice on acis data reduction and bing zhang for useful discussions of pulsar physics .
support for this work was provided by the nasa through chandra awards go1 - 2071x and go2 - 3091x , issued by the chandra x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of nasa under contract nas8 - 39073 .
this work was also partially supported by nasa grant nag5 - 10865 .
aschenbach , b. , & brinkmann , w. 1975 , a&a , 41 , 147 akhiezer , a. i. , akhiezer , i. a. , polovin , r. v. , sitenko , a. g. , & stepanov , k. n. 1975 , plasma electrodynamics , ( oxford : pergamon ) , p.112 begelman , m. c. 1998 , apj , 493 , 291 begelman , m. c. , blandford , r. d. , rees , m. j. 1984 , rev .
, 56 , 255 benford , g. 1984 , apj , 282 , 154 caraveo , p. a. , de luca , a. , mignani , r. p. , & bignami , g. f. 2001 , apj , 561 , 930 dodson , r. , legge , d. , reynolds , j. e. , & mcculloch , p. m. 2003a , apj , submitted ( astro - ph/0302374 ) dodson , r. , lewis , d. , mcconnel , d. , & deshpande , a. a. 2003b , mnras , submitted ( astro - ph/0302373 ) gaensler , b. m. , arons , j. , caspi , v. m. , pivovaroff , m. j. , kawai , n. , & tamura , k. 2002 , apj , 569 , 878 gallant , y. a. , & arons , j. 1994 , apj , 435 , 230 ginzburg , v. l. , & syrovatskii , s. i. 1965 , ara&a , 3 , 297 [ gs65 ] harding , a. k. , muslimov , a. g. , & zhang , b. 2002 , apj , 576 , 366 harnden , f. r. , grant , p. d. , seward , f. d. , & kahn , s. m. 1985 , apj , 299 , 828 helfand , d. j. , gotthelf , e. v. , & halpern , j. p. 2001
, apj , 556 , 380 hester , j. j. , mori , k. , burrows , d. , et al .
2002 , apj , 577 , l49 hibschman , j. , & arons , j. 2001 , apj , 560 , 871 kargaltsev , o. , pavlov , g. g. , sanwal , d. , & garmire , g. p. 2002
, in neutron stars in supernova remnants , eds .
p .. o. slane & b. m. gaensler , asp conf .
, v.271 ( san francisco : asp ) , 181 leahy , j. p. 1991 , in beams and jets in astrophysics , ed .
a. hughes ( cambridge : cambridge univ . press ) , 147 lewis , d. , dodson , r. , mcconnel , d. , & deshpande , a. 2002 , in neutron stars in supernova remnants , eds .
p .. o. slane & b. m. gaensler , asp conf .
ser . , v.271 ( san francisco : asp ) , 191 markwardt , c. b. , & gelman , h. 1998 , mem .
italiana , 69 , 927 mignani , r. p. , de luca , a. , kargaltsev , o. , pavlov , g. g. , zaggia , s. , caraveo , p. a. , & becker , w. 2003 , apj , submitted mori , k. , hester , j. j. , burrows , d. n. , pavlov , g. g. , & tsunemi , h. 2002 , in neutron stars in supernova remnants , asp conf .
, v. 271 , eds .
o. slane & b. m. gaensler ( san francisco : asp ) , 157 pacholczyk , a. g. 1970 , radio astrophysics ( san francisco : freeman ) pavlov , g. g. 2000 , aas head meeting 32 , # 07.05 pavlov , g. g. , sanwal , d. , garmire , g. p. , zavlin , v. e. , burwitz , v. , & dodson , r. g. 2000 , aas meeting 196 , # 37.04 pavlov , g. g. , kargaltsev , o. y. , sanwal , d. , & garmire , g. p. 2001a
, apj , 554 , l189 pavlov , g. g. , zavlin , v. e. , sanwal , d. , burwitz , v , .
& garmire , g. p. 2001b
, apj , 552 , l129 plucinsky , p. p. , et al .
2002 , in astronomical telescopes and instrumentation 2002 , eds .
j. e. trmper & h. d. tananbaum , spie conf .
, in press ( astro - ph/0209161 ) radhakrishnan , v. , & deshpande , a. 2000 , a&a , 379 , 551 townsley , l. k. , broos , p. s. , garmire , g. p. , & nousek , j. a. 2000 , apj , 534 , l139 zhang , l. , & cheng , k. s. 1997 , apj , 487 , 370 | observations of the vela pulsar - wind nebula ( pwn ) with the _ chandra x - ray observatory _ have revealed a complex , variable pwn structure , including inner and outer arcs , a jet in the direction of the pulsar s proper motion , and a counter - jet in the opposite direction , embedded in diffuse nebular emission .
the jet consists of a bright , @xmath0-long inner jet , between the pulsar and the outer arc , and a dim , curved outer jet that extends up to @xmath1 in approximately the same direction . from the analysis of thirteen _ chandra
_ observations spread over @xmath2 years we found that this outer jet shows particularly strong variability , changing its shape and brightness .
we observed bright blobs in the outer jet moving away from the pulsar with apparent speeds ( 0.3@xmath3 and fading on time - scales of days to weeks .
if the blobs are carried away by a flow along the jet , the observed variations suggest mildly relativistic flow velocities , about ( 0.3@xmath4 . the spectrum of the outer jet fits a power - law model with a photon index @xmath5 . for a distance of 300 pc ,
the apparent average luminosity of the outer jet in the 18 kev band is about @xmath6 erg s@xmath7 , compared to @xmath8 from the whole pwn within @xmath9 from the pulsar .
the x - ray emission of the outer jet can be interpreted as synchrotron radiation of ultrarelativistic electrons / positrons .
this interpretation allows one to estimate the magnetic field ,
@xmath10 @xmath11 g , maximum energy of x - ray emitting electrons , @xmath12 ev , and energy injection rate , @xmath13 , for the outer jet . in the summed pwn image , we see a faint , strongly bent , extension of the outer jet .
this bending could be caused by combined action of a wind within the supernova remnant , with a velocity of a few @xmath14 km s@xmath7 , along with the ram pressure due to the pulsar s proper motion .
the more extreme bends closer to the pulsar , as well as the apparent side motions of the outer jet , can be associated with kink instabilities of a magnetically confined , pinched jet flow .
another feature found in the summed image is a dim , @xmath15-long outer counter - jet , which also shows a power - law spectrum with @xmath161.5 . southwest of the jet / counter - jet ( i.e.
, approximately perpendicular to the direction of pulsar s proper motion ) , an extended region of diffuse emission is seen .
relativistic particles responsible for this radiation are apparently supplied by the outer jet . |
the multiple descriptions ( md ) problem has been studied extensively since late 1970s , see @xcite and the references therein . in a @xmath1descriptions md setup
, the encoder sends @xmath2 packets ( descriptions ) which are sent to the receiver over @xmath2 different channels . in the most general setting
, it is assumed that the decoder receives a subset of the descriptions without any error and the remaining are completely lost .
the decoder reconstructs the source upto a given level of distortion when a subset of the descriptions are received .
the goal of the md problem is to establish the complete rate - distortion region to trade - off the encoding rates to the achievable distortions .
the general setup has remained challenging and unsolved due to the intricacies of the problem in maintaining the balance between the full reconstruction quality versus quality of individual descriptions . until recently , for general sources and distortion measures , the most recognized achievable rate - distortion region for the @xmath1channel md setup was due to venkataramani , kramer and goyal ( vkg ) @xcite , whose encoding scheme builds on the prior work for the 2-channel case by el - gamal and cover ( ec ) @xcite and zhang and berger ( zb ) @xcite .
the vkg scheme involves a combinatorial number of refinement codebooks along with a single shared codebook used to control the redundancy across the descriptions .
we introduced a new encoding scheme in @xcite involving ` combinatorial message sharing ' ( cms ) which differs from the vkg scheme primarily in the number of shared codebooks .
the cms scheme allows for every subset of the descriptions to share a different common codebook , thereby leading to a combinatorial number of shared messages . at the time of submission of @xcite
, it was not known whether the cms scheme leads to a strictly improved rate - distortion region over the vkg scheme . in this paper ,
our objective is to prove by example that the new region is indeed strictly better .
specifically , we show that for a binary symmetric source under hamming distortion measure , the cms scheme achieves points outside the vkg region @xmath3 .
in fact , more generally , our result holds @xmath3 for any source and distortion measures for which the zb scheme achives points outside the ec scheme for the corresponding 2-descriptions problem .
we note in passing that , other encoding schemes have been proposed in the literature for certain special cases ( specific sources and distortion measures ) of the @xmath1channel md setup @xcite , which achieve points outside @xmath4 .
however , none of these schemes have been proven to subsume or outperform @xmath4 for general sources and distortion measures .
the potential implications of our results on these coding schemes are beyond the scope of this paper . in the following section , we formally state the @xmath1channel md setup and describe the prior results due to ec @xcite , zb @xcite , vkg @xcite and the cms scheme @xcite . in section [ sec :
proof - of - strict ] , we prove the strict improvement of the achievable region .
we follow the notation in @xcite .
a source produces @xmath5 iid copies , denoted by @xmath6 , of a generic random variable @xmath7 taking values in a finite alphabet @xmath8 .
we denote @xmath9 .
there are @xmath2 encoding functions , @xmath10 , which map @xmath11 to the descriptions @xmath12 , where @xmath13 for some @xmath14 .
the rate of description @xmath15 is defined as @xmath16 .
each of the descriptions are sent over a separate channel and are either received at the decoder error free or are completely lost .
there are @xmath17 decoding functions for each possible received combination of the descriptions @xmath18 , @xmath19 , where @xmath20 takes on values on a finite set @xmath21 , and @xmath22 denotes the null set . when a subset @xmath23 of the descriptions are received at the decoder , the distortion is measured as @xmath24 $ ] for some bounded distortion measures @xmath25 defined as @xmath26 .
we say that a rate - distortion tuple @xmath27 is achievable if there exit @xmath2 encoding functions with rates @xmath28 and @xmath17 decoding functions yielding distortions @xmath29 .
the closure of the set of all achievable rate - distortion tuples is defined as the ` _ _ @xmath2-channel multiple descriptions rd region _ _ ' .
note that , this region has @xmath30 dimensions . in what follows , @xmath31 denotes the set of all subsets ( power set ) of any set @xmath32 and @xmath33 denotes the set cardinality .
note that @xmath34 .
@xmath35 denotes the set complement . for two sets @xmath36 and @xmath37
, we denote the set difference by @xmath38 .
we use the shorthand @xmath39 for @xmath40 and @xmath41 .
@xmath39 is a set of variables , whereas @xmath41 is a single variable . ] . )
indicates ` common random variable ' .
red ( @xmath42 ) indicates ` base layer random variables ' and white ( @xmath41 @xmath43 ) indicates ` refinement random variables ' .
the arrow indicates the order of codebook generation.[fig : vkg ] ] the achievable region of @xcite is denoted here @xmath4 and is described as follows .
let @xmath44 be any set of @xmath45 random variables distributed jointly with @xmath7 .
then , an rd tuple is said to be achievable if there exist functions @xmath46 such that:@xmath47\label{eq : vkg}\end{aligned}\ ] ] @xmath48 .
the closure of the achievable tuples over all such @xmath45 random variables gives @xmath4 .
here , we only present an overview of the encoding scheme . the order of codebook generation of the auxiliary random variables is shown in figure [ fig : vkg ] .
first , @xmath49 codewords of @xmath50 are generated using the marginal distribution of @xmath50 .
conditioned on each codeword of @xmath50 , @xmath51 codewords of @xmath52 are generated according to their respective conditional densities .
next , for each @xmath53 , a single codeword is generated for @xmath54 conditioned on @xmath55 @xmath56 .
note that to generate the codebook for @xmath57 , we first need the codebooks for all @xmath58 and @xmath50 . on observing a typical sequence @xmath59 , the encoder tries to find a jointly typical codeword tuple one from each codebook .
codeword index of @xmath52 ( at rate @xmath60 ) is sent in description @xmath15 .
along with the ` _ _ private _ _ ' messages , each description also carries a ` _ _ shared message _ _ ' at rate @xmath61 , which is the codeword index of @xmath50
. hence the rate of each description is @xmath62 .
vkg showed that , to ensure finding a set of jointly typical codewords with the observed sequence , the rates must satisfy ( [ eq : vkg_rate ] ) .
it then follows from standard arguments ( see for example `` typical average lemma '' @xcite ) that , if the random variables also satisfy ( [ eq : vkg ] ) , then the distortion constraints are met .
note that , @xmath50 is the _ only _ shared random variable . @xmath63 form the base layer random variables and all @xmath64 form the refinement layers .
observe that the codebook generation follows the order : shared layer @xmath65 base layer @xmath65 refinement layer .
the vkg scheme for the 2-descriptions scenario involves 4 auxiliary random variables @xmath66 and @xmath67 .
the vkg region was originally derived as an extension of the ec @xcite and zb @xcite coding schemes , which were designed for the 2-descriptions scenario .
the first of the two regions was by el - gamal and cover and their rate region ( denoted here by @xmath68 ) is obtained by setting @xmath69 in @xmath4 , where @xmath70 is a constant .
zhang and berger ( their region is denoted here by @xmath71 ) later showed that , including the shared random variable can give strict improvement over @xmath68 .
their result , while perhaps counter - intuitive at first , clarifies the fact that , a shared message among the descriptions helps to better coordinate the messages , thereby providing a strictly improved rd region , even though it introduces redundancy .
we will describe their result in detail in section [ sec : proof - of - strict ] , as our example builds upon theirs . however , it is known that @xmath68 is complete for some special cases of the setup ( see for example @xcite ) . ] in this section , we briefly describe our cms encoding scheme in @xcite .
the vkg encoding scheme employs _
one _ common codeword ( @xmath50 ) that is sent in all the @xmath2 descriptions .
however , when dealing with @xmath72 descriptions , restricting to a single shared message could be suboptimal .
the cms scheme therefore allows for ` combinatorial message sharing ' , i.e a common codeword is sent in each ( non - empty ) subset of the descriptions . before describing the codebook generation and stating the theorem , we define the following subsets of @xmath73:@xmath74 let @xmath75 be any non - empty subset of @xmath76 with @xmath77 . we define the following subsets of @xmath78 and @xmath79:@xmath80 we also define to mean @xmath81 is subsumed in @xmath75 and @xmath82 to mean strictly subsumed.]:@xmath83 the shared random variables are denoted by ` @xmath84 ' .
the base and the refinement layer random variables are denoted by ` @xmath85 ' .
the codebook generation is done in an order as shown in figure [ fig : l_channel_cms ] .
first , the codebook for @xmath50 is generated .
then , the codebooks for @xmath86 , @xmath87 are generated in the order @xmath88 .
@xmath89 codewords of @xmath90 are independently generated conditioned on each codeword tuple of @xmath91 .
this is followed by the generation of the base layer codebooks , i.e. @xmath52 , @xmath92 .
conditioned on each codeword tuple of @xmath93 , @xmath51 codewords of @xmath52 are generated independently . then the codebooks for the refinement layers are formed by generating a single codeword for @xmath94 conditioned on every codeword tuple of @xmath95 .
observe that the base and the refinement layers in the cms scheme are similar to that in the vkg scheme , except that they are now generated conditioned on a subset of the shared codewords .
the encoder employs joint typicality encoding , i.e. , on observing a typical sequence @xmath59 , it tries to find a jointly typical codeword tuple , one from each codebook .
as with the vkg scheme , the codeword index of @xmath52 ( at rate @xmath60 ) is sent in description @xmath15 .
however , now the codeword index of @xmath86 ( at rate @xmath96 ) is sent in _ all _ the descriptions @xmath97 .
therefore the rate of description @xmath15 is:@xmath98 we next state the main result in @xcite which describes a new region for the @xmath1channel md setup achievable by the cms scheme .
let @xmath99 be any set of @xmath100 random variables jointly distributed with @xmath7 .
we define the quantities @xmath101 and @xmath102 as follows : @xmath103 we follow the convention @xmath104 .
let @xmath105 @xmath106 and @xmath60 @xmath107 be any set of rate tuples satisfying:@xmath108 then , the rd region for the @xmath1channel md problem contains the rates and distortions for which there exist functions @xmath46 , such that @xmath109\label{eq : dist_condition_thm}\end{aligned}\ ] ] the closure of the achievable tuples over all such @xmath100 random variables is denoted by @xmath110 .
observe that both the vkg and the cms schemes are same as the zb scheme for 2 descriptions scenario .
note that , the total number of auxiliary random variables in the cms scheme is almost twice that in the vkg scheme ( which already is exponential in @xmath2 ) . at the time of submission of @xcite , it was yet unclear if this increase pays off with an improved achievable region . the following theorem , being the main contribution of this paper , establishes that there exists scenarios for which @xmath110 is strictly larger than @xmath4 .
\(i ) the rate - distortion region achievable by the cms scheme is always at least as large as the region achievable by the vkg region , i.e.:@xmath111 ( ii ) there exists scenarios for which the cms scheme leads to a region strictly larger than that achievable by the vkg scheme , i.e.:@xmath112 specifically , for a binary symmetric source under hamming distortion measure , the cms scheme achieves a strictly larger rate - distortion region compared to the vkg scheme @xmath3 .
part ( i ) of the theorem is rather simple to prove and is a straight forward corollary of the main theorem in @xcite .
it follows directly by setting @xmath113 @xmath114 such that @xmath115 in @xmath110 .
we then have @xmath116 @xmath117 . substituting in ( [ eq : rate_condition_thm ] )
, we get @xmath118 which is same as ( [ eq : vkg ] ) .
we prove ( ii ) by considering the binary symmetric source example for which the cms scheme achieves points which can not be achieved by the vkg scheme .
note that , once we prove that the cms scheme achieves a strictly larger region for some @xmath119 , then it must be true for all @xmath120 .
hence to prove ( ii ) , it is sufficient for us to show that it is true for @xmath121 .
however , we first include an example for @xmath122 for building intuition and understanding of the type of scenarios where the cms scheme provides strict improvement
. then we will prove the result for @xmath121 .
we also note that obviously scenarios exit for which @xmath123 ( for example when @xmath4 is complete @xcite ) .
finding the set of all such scenarios is an interesting problem in itself and is beyond the scope of this paper . to describe our example ,
we require certain results pertaining to binary multiple descriptions @xcite and successive refinement of binary sources @xcite . in what follows , we state these results . *
the zhang - berger example * : zhang and berger proved that , for the binary symmetric 2-descriptions md problem under hamming distortion measure , sending a common codeword in both the descriptions provides a strict improvement over the ec scheme .
we briefly describe their result .
note that the rate - distortion region has 5 dimensions denoted by @xmath124 ) .
denote the rate - distortion region achievable by the ec scheme by @xmath125 and the corresponding region achievable by the zb scheme ( i.e. achieved by adding a common codeword among the two descriptions ) by @xmath126 .
obviously @xmath127 , as we can always choose not to send any common codeword in the zb scheme .
denote by @xmath128 the following cross section of @xmath125:@xmath129 similarly , denote by @xmath130 , the corresponding cross section of @xmath131 . to show that @xmath132 , they considered a particular joint probability mass function ( pmf ) @xmath133
let us denote the achievable region associated with this pmf by @xmath134 and the corresponding cross - section ( [ eq : ec_crosssection ] ) by @xmath135 .
they showed that @xmath136 such that @xmath137 .
we refer the reader to @xcite for a detailed derivation and the values of @xmath138 and @xmath139 .
* successive refinement * : the problem of successive refinement was first proposed by equitz and cover in @xcite and has since then been studied extensively by information theorists @xcite .
the problem is motivated by scalable coding , where the encoder generates two layers of information called the base layer and the enhancement layer .
the base layer provides a coarse reconstruction of the source , while the enhancement layer is used to ` refine ' the reconstruction beyond the base layer .
the objective is to encode the two layers such that the distortion at both the base and the enhancement layers are optimal .
this setup is shown schematically in figure [ fig : successive - refinement - setup ] .
observe that , the 2-layer successive refinement region is indeed a special case ( the cross - section @xmath140 ) of the 2-descriptions md setup where the distortion constraint on one of the individual descriptions is removed .
the complete rate region for successive refinement was derived in @xcite where it was shown that the ec coding scheme achieves the complete rate region .
an interesting followup question is that of ` _ _ successive refinability _ _ ' of sources .
assume @xmath141 , then a source is said to be successively refinable under @xmath142 if @xmath143 , the rate point @xmath144 is achievable , where @xmath145 denotes shannon s rate distortion function .
this condition implies that there is no loss in describing the source in two successive parts . an important point to note
is that , for a successively refinable source , when the encoder operates at @xmath146 , there is _ absolutely no redundancy _ between the two layers of information , i.e. , the _ two layers can not carry a common codeword_. we finally note that a binary symmetric source is successively refinable under hamming measure @xcite . * proof of ( ii ) : @xmath147 * : consider a 4-descriptions md problem for a binary symmetric source ( @xmath148 ) under hamming distortion measure .
the rate - distortion region consists of 19 dimensions .
we denote the region achievable by the vkg scheme by @xmath149 and that achievable using the cms scheme by @xmath150 .
we now consider a particular cross - section of these regions where we apply constraints only on @xmath151 and @xmath152 .
we remove the constraints on all other distortions , i.e. we set @xmath153 and @xmath154 to @xmath155 .
equivalently , we can think of a 4 descriptions md problem with a particular channel failure pattern , wherein only one of the following sets of descriptions can reach the decoder reliably : @xmath156 as shown in figure [ fig : example - to - demonstrate ] .
we denote the set of all achievable points for this setup using the vkg and the cms schemes by @xmath157 and @xmath158 respectively .
note that , this equivalent model is used simply for analysis purposes , while we are actually interested in a cross section of the general binary symmetric 4-descriptions region .
observe that , with respect to the first 2 descriptions , we have a simple 2-descriptions problem and with respect to the last 2 descriptions , we have a successive refinement problem . extending the arguments of zhang and berger , we define the following infimum of @xmath157 : @xmath159 denote the corresponding infimum of @xmath158 by @xmath160 .
recall that the vkg scheme forces all the descriptions to have a single common codeword .
constraints @xmath161 and @xmath162 ensure that descriptions 3 and 4 carry completely complementary information , i.e. they _ can not _ carry a common codeword .
is in fact redundant .
just the constraint @xmath162 is sufficient to establish that descriptions 3 and 4 can not carry a common codeword . however , as a binary source is successively refinable , we can always achieve @xmath163 and hence the constraint @xmath161 gets applied implicitly once we apply @xmath162 .
this implies that , the gains due to the cms scheme are not only restricted to successively refinable sources .
in fact , the cms scheme can achieve points outside the vkg region for any source and distortion measure for which the zb scheme achieves points outside the ec scheme for the corresponding 2-descriptions setup . ] as vkg coding scheme forces the _ same common codeword _
among all the 4 descriptions , it follows that:@xmath164 on the other hand , the cms scheme allows for distinct common codewords to be sent in each subset of the descriptions .
hence , we can send a common codeword only among the two descriptions 1 and 2 while still maintaining @xmath165 .
this is achieved by setting all the common random variables to @xmath70 except @xmath166 , which has joint pmf @xmath138 with @xmath167 .
this allows us to achieve : @xmath168 this implies that @xmath169 and hence @xmath170 .
this example clearly illustrates the freedom the cms scheme exhibits in controlling the redundancy across the messages . * @xmath171 * : in similar lines to the 4-descriptions case , we next consider a 3-descriptions md problem for a binary symmetric source under hamming distortion measure .
let the achievable regions be denote by @xmath172 and @xmath173 respectively .
we consider the cross - sections of the achievable regions where we apply constraints only on @xmath174 and @xmath175 as shown in figure [ fig:3d_eg ] .
we denote these cross - sections by @xmath176 and @xmath177 respectively .
now consider any point @xmath178 such that @xmath179 and @xmath180 , where for two sets @xmath81 and @xmath75 , @xmath181 . from the results of zhang and berger , if @xmath179 , descriptions 1 and 2 _ must _ carry a common codeword .
let the rate of the common codeword be @xmath182 .
vkg scheme forces this codeword to be sent as part of @xmath183 as well . as this common codeword
is received as part of both descriptions 1 and 3 , it is redundant in @xmath183 to achieve @xmath175 .
this implies that @xmath184 .
as there exit points in the boundary of @xmath172 which satisfy @xmath179 , the cms scheme achieves points outside the vkg scheme .
hence , we have shown that for a binary symmetric source under hamming distortion measure , the cms scheme achieves a strictly larger rate - distortion region than the vkg scheme for all @xmath72 .
we recently proposed a new encoding scheme for the general @xmath1channel multiple descriptions problem involving ` _ _ combinatorial message sharing _ _ ' ( cms ) which leads to a new achievable region subsuming the most well known region for this problem by venkataramani , kramer and goyal ( vkg ) @xcite for general sources and distortion measures . in this paper
, we showed that there exists scenarios ( particularly for a binary symmetric source under hamming distortion measure ) for which , the new region is strictly larger than that achievable by the vkg scheme .
as part of future work , we will investigate under what scenarios the cms scheme achieves the complete rd region .
j. wang , j. chen , l. zhao , p. cuff , and h. permuter , a random variable substitution lemma with applications to multiple description coding , preprint .
[ online ] .
available : http://arxiv.org/abs/0909.3135 .
k. viswanatha , e. akyol and k. rose , `` combinatorial message sharing for a refined multiple descriptions achievable region , '' to appear at the proceedings of ieee international symposium on information theory ( isit ) 2011 .
submitted version available at : http://www.scl.ece.ucsb.edu/kumar/isit_md_sub.pdf | we recently proposed a new coding scheme for the l - channel multiple descriptions ( md ) problem for general sources and distortion measures involving ` combinatorial message sharing ' ( cms ) @xcite leading to a new achievable rate - distortion region .
our objective in this paper is to establish that this coding scheme strictly subsumes the most popular region for this problem due to venkataramani , kramer and goyal ( vkg ) @xcite . in particular , we show that for a binary symmetric source under hamming distortion measure , the cms scheme provides a strictly larger region for all l@xmath0 .
the principle of the cms coding scheme is to include a common message in every subset of the descriptions , unlike the vkg scheme which sends a single common message in all the descriptions .
in essence , we show that allowing for a common codeword in every subset of descriptions provides better freedom in coordinating the messages which can be exploited constructively to achieve points outside the vkg region .
multiple descriptions coding , source coding , rate distortion theory |
the static properties of baryons constitute an important test for theoretical descriptions in the non - perturbative domain of qcd . in particular , different predictions for the magnetic moments of baryons were made in quark - models , chiral quark soliton models , and lattice qcd calculations ( see e.g. ref .
the experimental values for the magnetic moments of octet baryons ( @xmath5 ) are known accurately through spin precession measurements . for decuplet baryons , only the magnetic moment of the @xmath6
could be determined through such techniques , because the other decuplet baryons have too short lifetimes .
in particular the magnetic dipole moment of the @xmath1 resonance is of considerable theoretical interest .
if su(6 ) symmetry holds , then the nucleon and the @xmath3 resonance are degenerate and their magnetic moments are related through @xmath7 , where @xmath8 is the @xmath3 electric charge , and @xmath9 the proton magnetic moment .
however , different theoretical models predict considerable deviations from this su(6 ) value @xcite . in the past
, it has been proposed to determine the magnetic moment of the @xmath10 by measuring the @xmath11 reaction @xcite , and the first dedicated experiment of this type has been performed in the 70s @xcite . as a result of these measurements @xcite , and using different theoretical analyses , the pdg @xcite quotes the range : @xmath12 ( where @xmath13 is the nuclear magneton ) , while su(6 ) symmetry results in the value @xmath14 .
the large uncertainty in the extraction of @xmath15 from the data is due to large non - resonant contributions to the @xmath11 reaction because of bremsstrahlung from the charged pion ( @xmath16 ) and proton ( p ) . as an alternative method
, it has been proposed @xcite to determine the magnetic moment of the @xmath2 through measurement of the @xmath0 reaction .
first calculations for this reaction , including only the resonant mechanism , have been recently performed in refs .
@xcite . due to the small cross sections for this reaction , which is proportional to @xmath17 ,
a first measurement has only now been performed by the a2/taps collaboration at mami @xcite . in the near future ,
such experiments can be performed with much higher count rates by using @xmath18 detectors , such as planned e.g. by using the crystal ball detector at mami @xcite .
it is the aim of the present work to perform detailed calculations of the @xmath19 reaction , which include both resonant and non - resonant mechanisms , and to investigate in detail the sensitivity of this reaction to the @xmath2 magnetic dipole moment . in sec .
[ kinem ] , we start by specifying the kinematics and cross section of the @xmath0 reaction . in sec .
[ model ] , we describe in detail the effective lagrangian model for the @xmath0 reaction in the @xmath1 resonance region .
besides resonant mechanisms which contain the @xmath4 magnetic moment contribution , the model also includes the main non - resonant contributions to this reaction .
it is shown how the model for the @xmath20 process is obtained from an analogous model for the @xmath21 process by coupling a photon in a gauge invariant way to this latter process .
special attention is paid to the constraint which gauge invariance imposes on the electromagnetic coupling of a photon to a resonance of finite width , which has been discussed extensively in the literature in recent years for the analogous case of the @xmath22 gauge boson for which very accurate data have become available .
we then show in sec .
[ results ] , within the context of the developed effective lagrangian model , the results for the @xmath23 reaction , for both existing and forthcoming experiments . in particular , we study the sensitivity of both unpolarized cross sections and photon asymmetries to the @xmath4 magnetic dipole moment . finally , we present our conclusions and an outlook in sec .
[ conclusions ] .
in the @xmath0 process , a photon ( with four - momentum @xmath25 and helicity @xmath26 ) scatters on a proton target ( with four - momentum @xmath27 , and spin projection @xmath28 ) , and a photon ( with four - momentum @xmath29 and helicity @xmath30 ) , a @xmath31 ( with four - momentum @xmath32 ) and a proton ( with four - momentum @xmath33 , spin projection @xmath34 ) can be observed in the final state .
the kinematics of the @xmath0 reaction is characterized by 5 variables .
first , the total c.m .
energy squared is given by the usual mandelstam invariant @xmath35 .
equivalently to @xmath36 , we will use the energy of the intial photon @xmath37 in the c.m .
system of the intial @xmath38 system ( total c.m .
furthermore , there are two variables which characterize the photon in the total c.m
. system : its energy @xmath39 and polar angle @xmath40 respectively .
finally , two more variables are needed to specify the kinematics of the 3-body final state completely . for convenience ,
we choose them to be the pion angles in the rest frame of the @xmath41 system w.r.t .
the direction of the c.m .
momentum vector of the @xmath41 system : @xmath42 denotes the polar angle of the pion , and @xmath43 is its azimuthal angle . from these variables , one can reconstruct the invariant mass of the @xmath41 system as : @xmath44 and the pion momentum in the @xmath41 rest frame is calculated as : @xmath45 the unpolarized cross section for the @xmath0 reaction , which is differential w.r.t .
the outgoing photon energy and angles in the c.m . system , and differential w.r.t .
the pion angles in the @xmath41 rest frame is given by : @xmath46 where @xmath47 ( @xmath48 ) are the polarization vectors of the incoming ( outgoing ) photons respectively .
( [ eq : cross3b ] ) , @xmath49 is the tensor for the @xmath0 process which will be calculated in sec . [ model ] within the context of an effective lagrangian model .
further on , we will show results for partially integrated cross sections of the @xmath0 reaction , e.g. the cross section @xmath50 differential w.r.t .
the outgoing photon c.m .
energy , or the cross section @xmath51 differential w.r.t .
the photon c.m .
solid angle .
those are obtained by integrating the fully differential cross section of eq .
( [ eq : cross3b ] ) over the appropriate variables .
in order to calculate the resonant @xmath1 contribution to the @xmath0 reaction , we start from the @xmath1-resonance contribution to the @xmath52 process , corresponding to fig . [ fig : gap_piop_diag](a ) .
we describe the @xmath3-resonance in the rarita - schwinger formalism . it is well known that the description of a spin-3/2 field , such as the @xmath3-resonance , through a rarita - schwinger field , introduces an arbitrary parameter ( usually called @xmath53 in the literature , see e.g. @xcite ) .
this parameter arises when removing the spin-1/2 component of the vector - spinor rarita - schwinger field , through a contact transformation .
physical quantities , describing a spin-3/2 particle , should however be independent of this parameter @xmath53 and therefore care has to be taken in the calculations that the @xmath53-dependence of the @xmath3 vertices ( @xmath54 , @xmath55 ) and @xmath3 propagator cancel .
we follow here the procedure of refs .
@xcite , where it was shown that the @xmath53-dependent feynman rules involving the @xmath3 can be replaced by a set of @xmath53-independent ` reduced ' vertices and propagator . in this framework , the reduced @xmath3-propagator @xmath56 is given by @xcite : @xmath57 where @xmath58 and @xmath59 are the @xmath3 four - vector indices , @xmath60 is the four - momentum and @xmath61 the mass of the @xmath3 . to take account of the finite width of the @xmath3-resonance
, we follow the procedure of @xcite by using a complex pole description for the @xmath3-resonance excitation .
this amounts to the replacement : @xmath62 in the propagator of eq .
( [ eq : redprop ] ) . for the complex pole parameter of the @xmath1-resonance
, we use the values given by the pdg @xcite : @xmath61 = 1210 mev , @xmath63 = 100 mev .
further on , when calculating the @xmath3-resonance contribution to the @xmath0 process , we will discuss that this ` complex mass scheme ' guarantees electromagnetic gauge invariance , in contrast to using breit - wigner propagators with energy - dependent widths . with the propagator of eq .
( [ eq : redprop ] ) , the amplitude for the @xmath1-resonance contribution to the @xmath52 process , corresponding to fig .
[ fig : gap_piop_diag](a ) , is then given by : @xmath64 \ , n(p_n , s_n ) , \label{eq : delpi0}\end{aligned}\ ] ] where @xmath65 is the photon four - vector , @xmath66 is the proton electric charge ( @xmath67 = @xmath68 = 1/137 ) and @xmath69 the pion mass .
the @xmath55 coupling constant @xmath70 in eq .
( [ eq : delpi0 ] ) is taken from the @xmath71 decay , which yields : @xmath72 .
furthermore , in eq .
( [ eq : delpi0 ] ) , @xmath73 ( @xmath74 ) denote the magnetic ( electric ) @xmath54 vertices respectively , and are defined as : @xmath75 the magnetic @xmath54 coupling @xmath76 at the real photon point is taken from the maid00 analysis : @xmath77 = 3.02 @xcite .
the small electric @xmath54 coupling @xmath78 , corresponding to the @xmath79 ratio of about -2.5% for the @xmath80 transition @xcite , can be safely neglected for the sake of our subsequent analysis .
starting from the @xmath3-contribution of eq .
( [ eq : delpi0 ] ) to the @xmath52 process , we now construct the corresponding @xmath3-contribution to the @xmath0 process .
this is obtained by coupling a photon in a gauge invariant way to all charged particles in fig .
[ fig : gap_piop_diag](a ) . this way we obtain in the first place the diagrams with a photon attached to an external proton ( figs . [
fig : gap_piogap_diag](a1 ) and ( a3 ) ) .
their contributions to the tensor @xmath49 of eq .
( [ eq : cross3b ] ) is given by : @xmath81 \ , \nonumber \\
& \times & \ , { { \left ( \pindagger - \kpdagger + m_n \right ) } \over { -2 \ , p_n \cdot k ' } } \ , \left [ \gamma^\nu - \kappa_p \ , i \sigma^{\nu \rho } \ , { { k'_\rho } \over { 2 m_n } } \right ] \ , n(p_n ,
s_n ) , \\ { \mathcal m}^{\nu \mu}_{a3 } ( \gamma p \to \gamma \pi^0 p)\ , & = & \ , i \ , e^2 \ , { 2 \over 3 } \ , { { f_{\pi n \delta } } \over { m_\pi } } \ , ( p_\pi)^\alpha \ , \nonumber\\ & \times & \bar n(p_n^ { ' } , s'_n ) \ , \left [ \gamma^\nu - \kappa_p \ , i \sigma^{\nu \rho } \ , { { k'_\rho } \over { 2 m_n } } \right ] \ , { { \left ( \poutdagger + \kpdagger + m_n \right ) } \over { 2 \ , p_n^ { ' } \cdot k ' } } \,\tilde g_{\alpha \beta}(p_\delta ) \,\nonumber\\ & \times & \ , \left [ g_m \ , \gamma_m^{\beta \mu } \,+\ , g_e \ , \gamma_e^{\beta \mu } \right ] \ , n(p_n , s_n ) , \end{aligned}\ ] ] where we also included , besides the electric coupling of the emitted photon to the proton , the magnetic contribution ( @xmath82 is the proton anomalous magnetic moment ) .
in the soft - photon limit ( @xmath83 ) , the coupling of the photon to the external lines gives the only contribution .
however , at finite energy for the emitted photon , gauge invariance also requires the diagram with a photon attached to the intermediate @xmath4 ( fig.[fig : gap_piogap_diag](a2 ) ) , which corresponds to the tensor : @xmath84 \ , n(p_n , s_n),\end{aligned}\ ] ] where @xmath85 is the @xmath86 vertex : @xmath87 containing the photon coupling to a spin 3/2 field of a point - like particle ( i.e. @xmath88 ) @xcite , as well as the @xmath3 anomalous magnetic dipole moment contribution , which is proportional to @xmath89 .
the vertex for a point- spin-3/2 particle follows from the requirement of gauge invariance .
indeed , only the sum of the 3 diagrams fig .
[ fig : gap_piogap_diag](a1,a2,a3 ) is gauge - invariant , which is expressed through the electromagnetic ward identity relating the @xmath86 vertex and the @xmath3-reduced propagator : @xmath90 to include the finite width of the @xmath3 , the ` complex mass scheme ' was advocated in ref .
@xcite , i.e. the replacement of @xmath61 by the complex mass of eq .
( [ eq : complexmass ] ) _ everywhere _ in the propagator eq .
( [ eq : redprop ] ) , and not only in its denominator . in doing so
, it is easily checked that the ward identity of eq .
( [ eq : wardiden ] ) still holds . on the other hand , when using breit - wigner propagators , replacing @xmath91 in the denominator of eq .
( [ eq : redprop ] ) by @xmath92 , and with @xmath93 an energy dependent width , the @xmath3 propagators before and after the emission of the photon have different widths ( except in the soft - photon limit @xmath83 , where the contribution of diagram fig . [ fig : gap_piogap_diag](a2 ) vanishes relative to those of fig . [
fig : gap_piogap_diag](a1 ) and ( a3 ) ) .
therefore , when using energy - dependent widths at finite energy of the emitted photon , it is not possible to maintain the ward identity of eq .
( [ eq : wardiden ] ) with the point vertex ( @xmath88 ) of eq .
( [ eq : gadeldel ] ) . in recent years
, the issue of a gauge invariant definition of resonant masses , widths and partial widths has newly received attention in view of high precision measurements of the properties of the @xmath94 and @xmath95 gauge bosons at lep @xcite . in particular
, it has been shown in refs .
@xcite that a complex pole definition of a resonance provides a systematic way to maintain gauge invariance of the amplitude at any order of perturbation theory .
in particular , one - loop corrections to @xmath96 propagators and vertices have been calculated through a fermion loop @xcite , and it has been shown in the limit of massless fermions ( leptons , quarks ) in the loop that the simplest prescription to obtain gauge invariant amplitudes in the presence of finite width effects is to perform a laurent expansion of the full amplitude around the complex pole position of the resonance @xcite .
a similar observation has been made in ref .
@xcite , where the @xmath97 vector meson contribution to the @xmath98 decay has been studied as a source to extract the magnetic dipole moment of the @xmath99 meson .
the absorptive one - loop corrections to the propagator and vertices of the charged @xmath99 vector meson were calculated in ref .
@xcite through a pion loop . in the limit of vanishing mass of the pions in the loop
, it was further shown that the ` complex mass scheme ' , replacing the @xmath99 mass by a complex parameter in the lowest order expressions for propagators and vertices , provides a gauge invariant description in the presence of the finite width of the @xmath99 meson .
this ` complex mass scheme ' prescription was then applied in ref .
@xcite to the propagator and vertices of the @xmath100 resonance , in the description of the @xmath11 reaction .
it was argued for the @xmath3 resonance , similar to the spin-1 case , that the constraint imposed by gauge invariance leads to this complex mass scheme .
an explicit calculation of the absorption corrections to @xmath3 propagator and vertices through @xmath101 loops would be very valuable to check this conjecture . in this work
, we will follow the ` complex mass scheme ' prescription of eq .
( [ eq : complexmass ] ) , and apply it to the description of the @xmath4 resonance in the @xmath0 process .
this guarantees electromagnetic gauge invariance as expressed in eq .
( [ eq : wardiden ] ) .
besides the vertex for a point particle of spin 3/2 , which is constrained by gauge invariance , we also included in eq .
( [ eq : gadeldel ] ) the contribution of the @xmath3 anomalous magnetic dipole moment , proportional to @xmath89 .
this magnetic moment contribution is gauge invariant by itself .
the extraction of the value of @xmath89 is the principal motivation for constructing the present model to analyse the @xmath0 reaction .
the @xmath4 magnetic moment @xmath102 is then obtained from the @xmath4 anomalous magnetic moment @xmath103 as : @xmath104 where @xmath105 is the usual nuclear magneton .
the su(6 ) value predicts the value @xmath106 . assuming su(2 ) isospin symmetry , the pdg range for @xmath15 corresponds for @xmath4 to the range @xmath107 = 1.85 - 3.75 @xmath13 , which translates to the range @xmath108 = 1.43 - 3.92 .
in addition to the magnetic dipole moment ( m1 ) , the @xmath1 also has static monopole and quadrupole coulomb moments ( c0 , c2 ) and a magnetic octupole moment ( m3 ) .
it is obvious that the coulomb moment does not couple to the emitted photon , and the magnetic octupole moment is suppressed by two additional powers in the photon momentum .
there is , in principle , a contribution from electric ( transverse ) quadrupole radiation ( e2 ) .
however this is strongly suppressed by the fact that e2 radiation vanishes by time - reversal invariance if the initial and final states are exactly the same .
in fact , such radiation can only contribute by off - shell and recoil effects of the intermediate resonance .
in view of the small ratio @xmath109 for the @xmath110 transition , it is therefore ( unfortunately ! ) very unlikely that one will ever learn about the quadrupole moment of the @xmath1 by radiative pion photoproduction . in the present description ,
we therefore keep only the magnetic dipole moment of the @xmath1 .
although the @xmath52 process in the @xmath1 resonance region is dominated by the @xmath3-excitation process , it is well known that an accurate description also necessitates the inclusion of non - resonant mechanisms . in the following ,
we discuss such non - resonant mechanisms and calculate them for the @xmath112 process . at the high energy side of the @xmath3(1232)-region , the prominent non - resonant contribution to the @xmath52 reaction comes from vector - meson exchange , dominantly @xmath111-exchange as shown in fig .
[ fig : gap_piop_diag](b ) .
the amplitude of the @xmath111-exchange contribution is given by : @xmath113 \ , n(p_n , s_n ) , \end{aligned}\ ] ] where @xmath114 gev is the @xmath111-meson mass .
the @xmath115 coupling @xmath116 is well known from the radiative decay @xmath117 , which leads to the value : @xmath116 = 0.314 .
the @xmath118 coupling @xmath119 is taken from a high - energy analysis of @xmath120 : @xmath119 = 15 @xcite .
the ( isoscalar ) tensor coupling @xmath121 is generally found to be small and is set equal to zero in the present analysis , i.e. @xmath122 . when coupling a photon in a gaugeinvariant way to both charged proton lines in fig .
[ fig : gap_piop_diag](b ) , one obtains the corresponding diagrams of fig .
[ fig : gap_piogap_diag](b1 ) and ( b2 ) for the @xmath0 reaction .
their contributions to the tensor @xmath49 of eq .
( [ eq : cross3b ] ) are given by : @xmath123 \,\nonumber\\ & \times&\ , { { \left ( \pindagger - \kpdagger + m_n \right ) } \over { -2 \ , p_n \cdot k ' } } \ , \left [ \gamma^\nu - \kappa_p \
, i \sigma^{\nu \rho } \ , { { k'_\rho } \over { 2 m_n } } \right ] \ , n(p_n , s_n ) , \\ { \mathcal m}^{\nu \mu}_{b2 } ( \gamma p \to \gamma \pi^0 p)\,&=&\ , - i e^2 \ , g_{\omega nn } \ , { { g_{\omega \pi \gamma } } \over { m_\pi } } \ , { 1 \over { t - m_\omega^2 } } \ , \varepsilon^{\lambda \mu \alpha \beta } \ , k_\lambda \ , ( k - p_\pi)_\alpha \ , \nonumber\\ & \times & \ , \bar n(p_n^ { ' } , s'_n ) \ , \left [ \gamma^\nu - \kappa_p \
, i \sigma^{\nu \rho } \ , { { k'_\rho } \over { 2 m_n } } \right ] \ , { { \left ( \poutdagger + \kpdagger + m_n \right ) } \over { 2 \ , p_n^ { ' } \cdot k ' } } \ , \nonumber\\ & \times & \ , \left [ \gamma_\beta + \kappa_\omega \ , i \sigma_{\beta \sigma } \ , { { ( k - p_\pi)^\sigma } \over { 2 m_n } } \right ] \ , n(p_n , s_n ) , \end{aligned}\ ] ] at the low energy side of the @xmath1-region , the prominent non - resonant contribution to the @xmath52 reaction comes from the born contributions as shown in fig .
[ fig : gap_piop_diag](c1 ) and ( c2 ) .
they are evaluated here using pseudo - vector @xmath124 coupling , which yields the amplitudes : @xmath125 \ , n(p_n , s_n ) , \\
\varepsilon_\mu(k , \lambda ) \ ; { \mathcal m}^{\mu}_{c2 } ( \gamma p \to \pi^0 p)\ , & = & - \ , e \ , { { f_{\pi n n } } \over { m_\pi } } \ ; \varepsilon_\mu(k , \lambda ) \ ; \bar n(p_n^ { ' } , s'_n ) \ , \left [ \gamma^\mu + \kappa_p \ , i \sigma^{\mu \rho } \ , { { k_\rho } \over { 2 m_n } } \right ] \nonumber\\ & \times & \ , { { \left ( \poutdagger - \kdagger + m_n \right ) } \over { -2 \ , p_n^ { ' } \cdot k } } \ppidagger \ , \gamma_5 \ , n(p_n , s_n ) \ , . \end{aligned}\ ] ] the @xmath124 coupling constant is given by its well known value : @xmath126 @xcite .
the corresponding contributions to the @xmath0 reaction are obtained by coupling a photon in a gauge invariant way to all proton lines in fig .
[ fig : gap_piop_diag](c1 ) and ( c2 ) .
this yields the six diagrams of fig .
[ fig : gap_piogap_diag](c1-c6 ) whose contributions to the tensor @xmath49 of eq .
( [ eq : cross3b ] ) are given by : @xmath127 \ , { { \left ( \pindagger - \kpdagger + m_n \right ) } \over { -2 \ , p_n \cdot k ' } } \ , \left [ \gamma^\nu - \kappa_p \ , i \sigma^{\nu \sigma } \ , { { k'_\sigma } \over { 2 m_n } } \right ] \ , n(p_n ,
s_n ) , \label{eq : born1 } \\ & & { \mathcal m}^{\nu \mu}_{c2 } ( \gamma p \to \gamma \pi^0 p)\ , = - \ , e^2 \ , { { f_{\pi n n } } \over { m_\pi } } \ ; \bar n(p_n^ { ' } , s'_n ) \,\ppidagger \ , \gamma_5 \ ; { { \left ( \pindagger + \kdagger - \kpdagger + m_n \right ) } \over { 2 \ , p_n \cdot ( k - k ' ) \,-\ , 2 \ , k \cdot k ' } } \nonumber\\ & & \hspace{2 cm } \times \ , \left [ \gamma^\nu - \kappa_p \ , i \sigma^{\nu \sigma } \ , { { k'_\sigma } \over { 2 m_n } } \right ] \ , { { \left ( \pindagger + \kdagger + m_n \right ) } \over { 2 \ , p_n \cdot k } } \ , \left [ \gamma^\mu + \kappa_p \ , i \sigma^{\mu \rho } \ , { { k_\rho } \over { 2 m_n } } \right ] \ , n(p_n ,
s_n ) , \label{eq : born2 } \\ & & { \mathcal m}^{\nu \mu}_{c3 } ( \gamma p \to \gamma \pi^0 p)\ , = - \ , e^2 \ , { { f_{\pi n n } } \over { m_\pi } } \ ; \bar n(p_n^ { ' } , s'_n ) \ , \left [ \gamma^\nu - \kappa_p \ , i \sigma^{\nu \sigma } \ , { { k'_\sigma } \over { 2 m_n } } \right ] \ , { { \left ( \poutdagger + \kpdagger + m_n \right ) } \over { 2 \ , p_n^ { ' } \cdot k ' } } \ , \nonumber\\ & & \hspace{2 cm } \times \ , \ppidagger \ , \gamma_5 \ ; { { \left ( \pindagger + \kdagger + m_n \right ) } \over { 2 \ , p_n \cdot k } } \ , \left [ \gamma^\mu + \kappa_p \ , i \sigma^{\mu \rho } \ , { { k_\rho } \over { 2 m_n } } \right ] \ , n(p_n , s_n ) , \label{eq : born3 } \\ & & { \mathcal m}^{\nu \mu}_{c4 } ( \gamma p \to \gamma \pi^0 p)\ , = - \ , e^2 \ , { { f_{\pi n n } } \over { m_\pi } } \ ; \bar n(p_n^ { ' } , s'_n ) \ , \left [ \gamma^\mu + \kappa_p \ , i \sigma^{\mu \rho } \ , { { k_\rho } \over { 2 m_n } } \right ] \ , { { \left ( \poutdagger - \kdagger + m_n \right ) } \over { -2 \ , p_n^ { ' } \cdot k } } \nonumber\\ & & \hspace{2 cm } \times\ , \ppidagger \ , \gamma_5 \ ; { { \left ( \pindagger - \kpdagger + m_n \right ) } \over { -2 \ , p_n \cdot k ' } } \ , \left [ \gamma^\nu - \kappa_p \ , i \sigma^{\nu \sigma } \ , { { k'_\sigma } \over { 2 m_n } } \right ] \ , n(p_n ,
s_n ) , \label{eq : born4 } \\ & & { \mathcal m}^{\nu \mu}_{c5 } ( \gamma p \to \gamma \pi^0 p)\ , = - \ , e^2 \ , { { f_{\pi n n } } \over { m_\pi } } \ ; \bar n(p_n^ { ' } , s'_n ) \ , \left [ \gamma^\mu + \kappa_p \ , i \sigma^{\mu \rho } \ , { { k_\rho } \over { 2 m_n } } \right ] \ , { { \left ( \poutdagger - \kdagger + m_n \right ) } \over { -2 \ , p_n^ { ' } \cdot k } } \nonumber\\ & & \hspace{2 cm } \times \ , \left [ \gamma^\nu - \kappa_p \ , i \sigma^{\nu \sigma } \ , { { k'_\sigma } \over { 2 m_n } } \right ] \ , { { \left ( \poutdagger + \kpdagger - \kdagger + m_n \right ) } \over { -2 \ , p_n^ { ' } \cdot ( k - k ' ) \,-\ , 2 \ , k \cdot k ' } } \ , \ppidagger \ , \gamma_5 \ , n(p_n , s_n ) , \label{eq : born5 } \\ & & { \mathcal m}^{\nu \mu}_{c6 } ( \gamma p \to \gamma \pi^0 p)\ , = - \ , e^2 \ , { { f_{\pi n n } } \over { m_\pi } } \ ; \bar n(p_n^ { ' } , s'_n ) \ , \left [ \gamma^\nu - \kappa_p \ , i \sigma^{\nu \sigma } \ , { { k'_\sigma } \over { 2 m_n } } \right ] \ , { { \left ( \poutdagger + \kpdagger + m_n \right ) } \over { 2 \ , p_n^ { ' } \cdot k ' } } \ , \nonumber\\ & & \hspace{2 cm } \times \ , \left [ \gamma^\mu + \kappa_p \ , i \sigma^{\mu \rho } \ , { { k_\rho } \over { 2 m_n } } \right ] \ , { { \left ( \poutdagger - \kdagger + \kpdagger + m_n \right ) } \over { -2 \ , p_n^ { ' } \cdot ( k - k ' ) \,-\ , 2 \ , k \cdot k ' } } \ , \ppidagger \ , \gamma_5 \ , n(p_n , s_n ) \ , .
\label{eq : born6}\end{aligned}\ ] ] note that all six amplitudes of eqs.([eq : born1]-[eq : born6 ] ) are needed to satisfy gauge invariance with respect to both the incoming and outgoing photons .
having described the model for the @xmath0 reaction , we now turn to the results . in order to have a point of comparison , we start by showing the result for the @xmath52 reaction corresponding to the mechanisms of fig . [
fig : gap_piop_diag ] .
our aim here is not to provide an accurate description of the @xmath52 reaction ( which exists in the literature ) , but to see the intrinsic accuracy of the present effective lagrangian formalism , which serves as our starting point for the corresponding mechanisms as shown in fig .
[ fig : gap_piogap_diag ] for the @xmath112 reaction .
[ fig : tot_ppio ] shows the total cross section for the @xmath21 reaction .
although the @xmath52 cross section is clearly dominated by the @xmath3-excitation mechanism , making it an informative tool to study the @xmath3-resonance properties , additional non - resonant mechanisms are needed to obtain an accurate description .
it is seen that at the lower energies , the main non - resonant contributions come from the nucleon born diagrams . at the higher energies , the @xmath3-excitation alone yields too large cross sections and
is mainly reduced by the @xmath111-exchange mechanism , which dominates the high - energy behaviour of the @xmath52 reaction @xcite . by comparing the total result in fig .
[ fig : tot_ppio ] with the accurate data , it is seen that the present effective lagrangian model provides a rather good description below the @xmath3-resonance position , underestimates the data around 325 mev at the 10 % level , and has an accuracy at the 20 % level around 450 mev .
the systematic deviation of the present tree - level description with regard to the data as one moves through the @xmath3-resonance , is well understood and originates from @xmath101 rescattering ( loop ) contributions which restore unitarity .
although below two - pion production threshold , one can easily restore unitarity for the @xmath52 reaction by using the @xmath101 phase - shifts as phases of the pion photoproduction amplitude , the corresponding unitarization procedure for the @xmath0 reaction has not yet been worked out . in particular , for the @xmath0 reaction , care has to be taken to respect gauge invariance , as one can have intermediate @xmath128 loop contributions where the photon couples to the intermediate charged pion .
such a gauge invariant unitary description of the @xmath0 reaction could be addressed within the context of a dynamical model , and seems to be a promising subject for future work . in the present paper , we explore as a first step the @xmath0 reaction within the context of the effective lagrangian model of fig . [
fig : gap_piogap_diag ] , in order to investigate promising kinematics and observables for the extraction of the @xmath4 magnetic dipole moment .
passing next from the @xmath52 reaction to the description of the @xmath0 reaction , it is important to stress that the electric coupling of the emitted photon to the charged particles is completely constrained by gauge invariance as discussed in sec . [ model ] .
however the magnetic dipole moment of the @xmath4 ( and also its smaller higher moments ) is not fixed by general theoretical arguments and has to be extracted from a fit to experiment .
therefore , the only new parameter that appears in the present formalism for the @xmath0 reaction compared to the @xmath52 reaction , is the @xmath4 anomalous magnetic moment @xmath108 . in the following ,
we investigate the sensitivity of different @xmath0 observables to this anomalous magnetic moment @xmath108 .
[ fig : gap_ppioga_en1 ] shows the dependence of the differential cross section of the @xmath0 reaction on the outgoing photon energy for three incoming photon energies through the @xmath3-resonance region .
the cross sections have been integrated over both photon and pion angles .
the calculations compare the importance of the different mechanisms in fig .
[ fig : gap_piogap_diag ] , using the value @xmath108 = 3 .
one clearly sees that the cross section for the resonant mechanism alone is dominated at the lower photon energies by the bremsstrahlung contributions ( fig .
[ fig : gap_piogap_diag](a1 ) and ( a3 ) ) , which show the typical 1/@xmath129 energy dependence .
it is also seen that the non - resonant mechanisms induce , due to interference , a distinct energy dependence as one moves through the @xmath3-resonance similar to the case of the @xmath130 reaction .
note in particular the partial cancellation of bremsstrahlung with @xmath3 and nucleon intermediate states for the smaller values of @xmath129 .
the cross section for the @xmath0 reaction has been measured for the first time by the a2/taps collaboration at mami @xcite .
our total results in fig .
[ fig : gap_ppioga_en1 ] compare favorably with the preliminary data in both incoming and outgoing photon energy dependence .
[ fig : gap_ppioga_en2 ] shows the sensitivity of the angle integrated cross sections to the value of the @xmath4 anomalous magnetic moment .
the sensitivity of these cross sections to @xmath108 increases with increasing outgoing photon energy , but remains relatively modest even at the higher energies , where the difference between the @xmath131 and @xmath132 values is at the 10 - 15 % level . in fig .
[ fig : gap_ppioga_ang_taps ] , we show the corresponding photon c.m . angular dependence of the cross sections integrated over the outgoing photon energy ( for the cut @xmath133 mev ) and over the pion angles .
the angular dependence displays a broad maximum around a c.m .
angle of 110@xmath134 .
the calculations show that the sensitivity of these integrated unpolarized cross sections to @xmath108 lies at the 10 - 20 % level over most of the angular range . at an incoming photon energy of 450 mev ,
a larger effect is observed around the forward and backward angles .
however , we found that the sensitivity to the @xmath4 anomalous magnetic moment can be substantially increased when using a linearly polarized photon beam . in this case
, one can define the photon asymmetry @xmath135 , where @xmath136 ( @xmath137 ) are the differential cross sections for incoming photon polarization perpendicular ( parallel ) to the reaction plane spanned by the incoming and outgoing photon c.m .
momentum vectors . the outgoing photon energy dependence of the photon asymmetry is shown in fig .
[ fig : gap_ppioga_en_asymm ] where the differential cross sections @xmath136 and @xmath138 are integrated over all photon and pion angles , corresponding to the situation in fig .
[ fig : gap_ppioga_en2 ] . in comparison with the unpolarized cross section shown in fig .
[ fig : gap_ppioga_en2 ] , one now obtains a clearly larger sensitivity to @xmath139 , in particular at the larger values of @xmath140 . one can further increase the sensitivity of the @xmath20 reaction to the @xmath4 magnetic dipole moment by measuring differential cross sections .
in particular , by chosing those angular ranges where the photon is emitted in the backward hemisphere with respect to both protons ( in the c.m .
system ) as shown in fig .
[ fig : kinem_forw ] , one suppresses the bremsstrahlung contributions and largely increases the sensitivity to @xmath108 .
this is shown in fig .
[ fig : diffasymm4 ] for the fully differential cross section of eq .
( [ eq : cross3b ] ) , which is differential w.r.t .
the photon c.m .
energy , the photon c.m .
polar angle and the pion polar and azimuthal angles ( the latter two are defined in the @xmath41 rest frame ) .
the differential cross sections in fig .
[ fig : diffasymm4 ] , shown for @xmath141 = 450 mev , are now peaked around @xmath140 = 100 mev due to the resonant-@xmath3 contribution of fig .
[ fig : gap_piogap_diag](a2 ) , which contains @xmath108 . on the other hand , in the angle integrated cross sections shown in fig .
[ fig : gap_ppioga_en2 ] , the contribution of fig .
[ fig : gap_piogap_diag](a2 ) merely manifests itself in the shoulder at higher values of the outgoing photon energy @xmath140 .
furthermore , one sees from fig .
[ fig : diffasymm4 ] that the maximum of the cross section for @xmath131 is reduced by about 30 % for the value @xmath132 as compared to @xmath131 .
the corresponding photon asymmetry @xmath142 reaches large values and also displays a very pronounced energy dependence .
the differences resulting from various values of @xmath108 is in the range predicted by different theoretical models , and should be clearly resolved by measurements of @xmath142 at the 5 % level .
although a measurement of the fivefold differential cross section is very tough due to the small count rates , we found that one can largely keep the sensitivity to @xmath89 by integrating over a larger region of phase space around the kinematical situations where the photon is emitted in the backward hemisphere with respect to both protons .
this is illustrated in fig .
[ fig : gap_ppioga_en3 ] where the differential cross section is shown integrated over the forward hemisphere for the outgoing photon angle ( w.r.t .
the incoming photon direction in the c.m .
system ) and where the pion is emitted over the azimuthal angular range @xmath143 .
one sees from fig .
[ fig : gap_ppioga_en3 ] that the cross section at 450 mev shows a 20 % sensitivity , and at 500 mev a 30 % sensitivity , to the difference between @xmath131 and @xmath132 . furthermore , by comparing for a value @xmath132 , the full result with the @xmath3 anomalous magnetic moment contribution separately ( diagram fig .
[ fig : gap_piogap_diag ] ( a2 ) ) , one sees that the latter yields the dominant contribution to the total result at the higher outgoing photon energies for 450 mev and 500 mev initial photon energies . by comparing the partially integrated cross section of fig .
[ fig : gap_ppioga_en3 ] at 450 mev to the cross section of fig .
[ fig : gap_ppioga_en2 ] , one sees that it reaches about one fifth of the cross section integrated over the full angular range .
as the cross section of fig .
[ fig : gap_ppioga_en2 ] has already been measured @xcite , the cross sections of fig .
[ fig : gap_ppioga_en3 ] could be accessible through a dedicated experiment . in view of the characteristic angular dependence of the effect under investigation
, a full @xmath18 coverage will be a prerogative to obtain more significant information on the magnetic dipole moment of the @xmath3 .
such a @xmath18 detector will allow one to increase the count rate substantially w.r.t .
the present taps experiment , and thus to select kinematical ranges where background ( bremsstrahlung ) effects are strongly suppressed as is shown in fig .
[ fig : gap_ppioga_en3 ] .
finally , we show in fig .
[ fig : gap_ppioga_en3_asymm ] the photon asymmetries corresponding to the partially integrated cross sections of fig .
[ fig : gap_ppioga_en3 ] . compared to the asymmetries of fig .
[ fig : gap_ppioga_en_asymm ] , where the cross sections were integrated over all photon and pion angles , both the magnitude of the asymmetries in fig .
[ fig : gap_ppioga_en3_asymm ] and the sensitivity to @xmath89 are larger .
therefore , a dedicated experiment consisting of a @xmath18 detector and a high intensity polarized photon beam , promises to provide a good opportunity to extract the @xmath4 magnetic dipole moment from the @xmath0 reaction .
such an experiment is planned using the crystal ball detector in combination with the polarized photon beam at mami @xcite .
in this paper we investigated the @xmath0 process as a tool to access the @xmath2 magnetic dipole moment .
an effective lagrangian formalism to describe the @xmath52 reaction in the @xmath1 resonance region , served as a starting point to develop a corresponding model of the @xmath0 reaction .
when extending the model of the @xmath52 process to the @xmath0 process , the electric coupling of the additional photon to the charged particles is completely constrained by gauge invariance .
in particular , we paid special attention to the constraint that gauge invariance imposes on the coupling of a photon to a resonance of finite width . in analogy to the case of the @xmath22 vector boson , which has been discussed extensively in recent years in the literature ,
we follow a complex mass scheme procedure where the amplitude for the @xmath0 process which involves the @xmath2 resonance , is laurent expanded around the complex pole position of the resonance .
this procedure guarantees electromagnetic gauge invariance . in comparison with the @xmath52 reaction ,
the only new parameter which enters in the description of the @xmath144 reaction in the @xmath1 region is the @xmath4 anomalous magnetic moment @xmath108 .
we then investigated the sensitivity of various @xmath0 observables to this anomalous magnetic moment @xmath108 .
we first compared the unpolarized @xmath0 cross section integrated over the whole photon and pion angular range .
for this fully angle integrated cross section , the resonant mechanism is dominated by the bremsstrahlung contribution which shows the typical @xmath145 dependence in the outgoing photon energy @xmath146 .
we also found that the non - resonant mechanisms induce , due to interference , a distinct energy pattern as one moves through the @xmath3-resonance .
our calculations compare favorably with preliminary data for those fully integrated cross sections .
the sensitivity of the unpolarized @xmath0 cross section integrated over the whole photon and pion angular range to @xmath108 , was found to increase with increasing photon energy , but remains relatively modest . at an incoming photon energy of 450 mev
, the difference between the @xmath131 and @xmath132 values is at the 10 - 15 % level .
however , one can further increase the sensitivity to @xmath108 by measuring differential cross sections .
in particular , by selecting the kinematical situations where the photon is emitted in the backward hemisphere w.r.t . both protons , the bremsstrahlung contribution to the cross sections is largely suppressed .
when partially integrating the cross sections around these backward angle kinematics , we found that , for an incoming photon energy range of 450 - 500 mev , the difference between the @xmath131 and @xmath132 values is at the 20 - 30 % level .
furthermore , we showed that the sensitivity to the @xmath3 anomalous magnetic moment can be even more increased by using a linearly polarized photon beam . at the higher outgoing photon energies , the asymmetries are at the 10 - 30 % level through the @xmath3 region and display a clear sensitivity to @xmath108 . in view of our findings , a dedicated experiment using a @xmath18 detector and a high intensity photon beam seems to be a promising experiment to extract the @xmath4 magnetic dipole moment from the @xmath0 reaction .
the model developed in this paper can serve as a first step for such an extraction .
the measurement of the @xmath3 magnetic dipole moment will provide us with new information on baryon structure which can be confronted with various baryon structure calculations .
it is a pleasure to acknowledge fruitful conversations with r. beck , m. kotulla , v. metag , b. nefkens , g. rosner , and l. tiator .
this work was supported by the deutsche forschungsgemeinschaft ( sfb 443 ) .
kondratyuk und l.a .
ponomarov , yad . fiz .
* 7 * , 11 ( 1968 ) [ sov .
* 7 * , 82 ( 1968 ) ] .
et al . _ ,
d * 18 * , 3911 ( 1978 ) .
a. bosshard et al .
d * 44 * , 1962 ( 1991 ) .
d.e . groom _
et al . _ ( pdg ) , eur .
j. * c15 * , 1 ( 2000 ) .
giannini , in _ proceedings of the workshop perspectives on nuclear physics at intermediate energies , ictp trieste , italy , 1983 ( world scientific , singapore , 1984 ) _ ; and d. drechsel , in mami funding proposal to dfg , sfb 201 ( 1984 - 86 ) , p. 56 .
machavariani , a. faessler , and a. j. buchmann , nucl . phys . *
a646 * , 231 ( 1999 ) ; nucl . phys . *
a686 * , 601 ( e ) ( 2001 ) .
d. drechsel , m. vanderhaeghen , m.m .
giannini , and e. santopinto , phys .
b * 484 * , 236 ( 2000 ) .
m. kotulla ( for the a2/taps collaboration ) , in _ proceedings of the workshop on the physics of excited nucleons ( nstar 2001 ) , mainz , germany , 2001 , ( world schientific , singapore , to be published)_. r. beck , b. nefkens _ et al .
_ , letter of intent , mami ( 2001 ) .
m. el amiri , g. lpez castro , and j. pestieau , nucl .
* a543 * , 673 ( 1992 ) . g. lpez castro and a. mariano , nucl - th/0010045 .
nath , b. etemadi , and j.d .
kimel , phys .
d * 3 * , 2153 ( 1971 ) .
d. drechsel , o. hanstein , s. kamalov , and l. tiator , nucl . phys .
* a645 * , 145 ( 1999 ) .
r. beck , h.p .
et al . _ ,
c * 61 * , 035204 ( 2000 ) .
a. sirlin , phys .
lett . * 67 * , 2127 ( 1991 ) ; phys .
86 * , 389 ( 2001 ) .
stuart , phys .
lett . * 70 * , 3193 ( 1993 ) .
h. veltman , z. phys .
c * 62 * , 35 ( 1994 ) .
g. lpez castro , j.l .
lucio m. , and j. pestieau , int .
a * 11 * , 563 ( 1996 ) ; hep - ph/9504351 .
u. baur and d. zeppenfeld , phys .
lett . * 75 * , 1002 ( 1995 ) .
m. beuthe , r. gonzalez felipe , g. lpez castro , and j. pestieau , nucl . phys .
* b498 * , 55 ( 1997 ) . g. lpez castro and g. toledo snchez , phys .
d * 61 * , 033007 ( 2000 ) . | the @xmath0 reaction in the @xmath1-resonance region is investigated as a method to access the @xmath2 magnetic dipole moment .
the calculations are performed within the context of an effective lagrangian model containing both the @xmath3-resonant mechanism and a background of non - resonant contributions to the @xmath0 reaction .
results are shown both for existing and forthcoming @xmath0 experiments .
in particular , the sensitivity of unpolarized cross sections and photon asymmetries to the @xmath4 magnetic dipole moment is displayed for those forthcoming data . |
an important aspect of strongly - correlated electron physics is the expectation that some systems may exhibit novel non - fermi liquid behavior .
this can present some interesting challenges for theoretical models since non - fermi liquid behavior is known to be rare based on simple renormalization group ( rg ) arguments@xcite .
the best understood example is the luttinger liquid in one spatial dimension , where the non - fermi liquid behavior is essentially attributed to the fact that quartic interactions of dirac fermions are marginal in the rg sense . in higher dimensions quartic interactions of dirac fermions are irrelevant .
these considerations were one of the primary initial motivations for the construction and analysis of a new model of interacting fermions@xcite with non - fermi liquid behavior . like the luttinger liquid
, the model is a continuum field theory of only 4 fermionic fields @xmath7 which carry charge and spin , with a unique quartic interaction due to fermi statistics .
the novelty of the model is the free kinetic term in the hamiltonian which in addition the commonplace term that is second order in spatial derivatives @xmath8 , contains an additional contribution that is second order in time derivatives @xmath9 , and thus has an emergent lorentz symmetry
. the model can be consistently canonically quantized as a fermion , even though this structure of the kinetic term is usually associated with relativistic bosons . in two spatial dimensions
the field @xmath10 has dimension @xmath11 , which implies the quartic interaction has dimension @xmath12 and is thus rg relevant .
the model in fact has a low - energy fixed point , i.e. quantum critical point .
several interesting properties in addition to the non - fermi liquid behavior emerged in the analysis of the model@xcite .
it has a hidden @xmath13 symmetry which contains the spin @xmath14 and electric @xmath15 as commuting subgroups .
the fundamental fields @xmath16 transform in the 4-dimensional spinor representation of @xmath13 and the bilinears decompose as @xmath17 .
the @xmath18 vector representation serves as order parameters for the spontaneous symmetry breaking of @xmath14 , i.e. magnetic order , and also contains cooper pair fields of charge @xmath19 which are order parameters for symmetry breaking of @xmath15 , i.e. superconductivity . by deriving separate gap equations and studying their solutions
it was shown that the model can contain an anti - ferromagnetic phase ( af ) and a d - wave superconducting phase ( sc ) . in this model the basic mechanism that leads to a d - wave sc instability is clearly identified as arising from the 1-loop scattering of cooper pairs .
an attractive feature of the model is its simplicity , reflected in the fact that it has very few free parameters : an overall energy scale @xmath20 set by the high - energy cutoff @xmath21 , a fermi velocity @xmath22 , and a single coupling @xmath23 .
furthermore , at low energies @xmath23 is near the low - energy fixed point value of @xmath24 , so that perturbation theory can be reliable .
the above features led us to propose this theory as a toy model of high temperature superconductivity ( htsc ) . here
the @xmath13 symmetry is quite different from the @xmath13 symmetry of zhang in this context@xcite : whereas he postulates an @xmath13 invariant landau - ginsburg theory for the bosonic spin and electric order parameters , our model is a microscopic model of fermions where the bosonic order parameters are composite bilinears in the fermions .
the @xmath13 symmetry was not put in by hand in our model , rather it is accidental and hidden .
in fact , the af and d - wave sc phases are _ not _ related by the @xmath13 in our model . in principle
a landau - ginsburg effective theory can be derived from our model , but like the bcs theory , our underlying microscopic theory is more powerful for deriving gap equations , critical temperatures , etc .
arguments that motivate its application to htsc were given in @xcite , the best one being its relation to the non - linear @xmath25 sigma model effective theory for the @xmath26 heisenberg model description of the af phase , which is also a relativistic theory ; however a rigorous derivation of it from say the hubbard model on the lattice is lacking .
if our model really turns out to correctly capture some essential features of htsc , then it should be viewed as an effective low - energy theory for wavelengths that are long compared to the lattice spacing .
one expects that such a description can be approximately described by a rotationally invariant quantum field theory describing the gas of quasi - particles that sc condenses out of .
the shortcoming of such a theory is that lattice effects , such as features that are dependent on the detailed structure of the fermi surface in the brillouin zone , are necessarily absent in the basic model , although perhaps they can be incorporated with small perturbations of the hamiltonian .
furthermore , our model does not have an explicit fermi surface , as for the @xmath25 sigma model of the af phase at half - filling .
nevertheless , the existence of a simple model of fermions with only repulsive quartic interactions that possesses the all of the main motifs of htsc , which is to say it has clearly identified mechanisms for the d - wave sc , the pseudogap , and anti - ferromagnetism , can be useful for deciding which properties are essential and which are superfluous for the phenomenon .
thus , although it has not been established that our model captures what is really happening in htsc , for the remainder of this paper we will take the liberty to borrow the terminology of the htsc literature where appropriate .
the main purpose of the present article is study in detail the origin and properties of the so - called pseudogap in our model . in the htsc materials
, the pseudogap refers to an energy scale @xmath27 where a cross - over behavior is observed in a variety of physical properties such as electronic specific heat , magnetic susceptibility and conductivity .
the same energy scale can be observed as the onset of a depression in the density of states . in the underdoped region , @xmath28 is considerably larger than the superconducting @xmath3 . for reviews
see @xcite .
the origin and physical interpretation of the pseudogap remains a fundamental question in the physics of htsc , and many researchers feel that a proper understanding of it will be an important key toward unraveling the mysteries of htsc .
two broad classes of theories can be summarized by the schematic phase diagrams shown in figure [ figure1 ] . in the scenario on the left
, @xmath28 goes to zero inside the sc dome , perhaps terminating at a quantum critical point . in this class of theories
the pseudogap is unrelated , and in fact competes with , superconductivity . in the second class of theories ,
the pseudogap indicates pre - formed cooper pairs for example , i.e. is a friendly precursor to sc , and the pseudogap line merges with @xmath3 on the overdoped side .
these opposing scenarios are discussed in some detail in @xcite .
thermodynamic data such as specific heat favors the first scenario@xcite , whereas spectroscopic data seems to support the second@xcite .
[ figure1 ] in our model there is an obvious identification of the pseudogap : whereas af or sc order are related to the order parameters in the @xmath18 vector of @xmath13 , the pseudogap is naturally associated with the singlet bilinear in the tensor product of @xmath29 , as suggested by tye@xcite , and also more tentatively in @xcite .
this corresponds to the operator @xmath30 .
whereas this term is classically zero in the relation with the @xmath25 sigma model , it is dynamically generated in the presence of interactions . as a contribution to the hamiltonian , it corresponds to a relativistic mass @xmath0 , and the 1-particle states have energy @xmath31 .
thus a non - zero mass @xmath0 gives a clean gap in the density of states , i.e. there is nothing `` pseudo '' about it .
as discussed above , the anisotropy of the observed pseudogap is not a feature of this mass - gap since it is a lattice effect ( `` fermi arcs '' ) . in the sequel
we identify this mass with the pseudogap energy scale @xmath32 and will show that its properties closely parallel the phenomenology of the pseudogap in htsc , at least on the underdoped side .
the pseudogap has an entirely different origin than the sc gap in our model , essentially because it corresponds to a dynamically generated vacuum expectation value @xmath33 which preserves the @xmath13 symmetry whereas sc breaks the electric @xmath15 subgroup .
it is an intrinsic property of the normal state density of states . whether it competes with or aids superconductivity
is straightforwardly addressed by incorporating it into the d - wave gap equation derived in @xcite .
as we will see , it clearly competes with superconductivity since if it is too large it destroys the solution to the sc gap equation .
fortunately , the pseudogap goes to zero at the critical point and this makes superconductivity possible ; in fact sc is enhanced near the critical point .
thus our model is in the class of the left figure in figure [ figure1 ] , with @xmath28 terminating at a quantum critical point near optimal doping .
( the proper treatment of temperature presented in this paper led to the identification of the critical point which differs from the original proposal in @xcite . ) our analysis of the pseudogap required two technical improvements of the work presented in @xcite .
in the latter work , temperature was treated crudely as a mass , the idea being that temperature and a mass can both be viewed as an infra - red cutoff ; this is ultimately unreliable for the computation of thermodynamic properties . in the present work ,
since the mass is identified with the pseudogap which is a zero temperature property , temperature must be dealt with properly , and this is accomplished with the matsubara formalism .
a non - zero mass also resolves some infra - red divergences that were present in the previous treatment .
the remainder of this article is organized as follows . in the next section we review the definition of the model and some of its basic properties .
section iii addresses the unitarity of the model , beyond what is contained in our previous work .
section iv describes our rg prescriptions . in section
v we analyze the dynamical generation of a mass , i.e. pseudogap , by deriving a gap equation for the vacuum expectation value @xmath34 .
the pseudogap depends on the variable @xmath35 which up to a scale is the inverse coupling . in section
vi we propose a relation between the variable @xmath35 and hole doping which thus gives the doping dependence of the pseudogap . in section vii
we compute the effect of the pseudogap on the electronic specific heat .
a non - zero magnetic field is introduced in section viii , and we compute the temperature dependent spin susceptibility and the magnetic field dependence of the specific heat . for all of these thermodynamic properties ,
we find crossover behavior at the temperature @xmath36 and its qualitative dependence on doping compares favorably with data . finally in section ix
we derive the finite temperature version of the d - wave gap equation and incorporate the effect of the pseudogap into it .
analysis of this equation clearly shows that the pseudogap competes with sc , and leads to a computation of the superconducting @xmath3 .
as for any second - quantized description of electrons with spin @xmath37 , the fundamental fields of the model are 4 fermionic fields @xmath38 , where the flavor index @xmath39 corresponds to spin and @xmath40 is electric charge . due to the fermionic statistics there is a unique quartic interaction , thus various models are primarily characterized by the free kinetic term .
our model in two spatial dimenions is [ ham ] h = d^2 @xmath41 the above hamiltonian would be a standard second - quantized field theory for fermions interacting via a delta - function potential if it were nt for the term that is second order in time derivatives , and this is the primary novelty of the model .
this choice of kinetic term can be motivated from the phenomenology of htsc , since it leads to the correct temperature dependence of the specific heat @xmath42 at low temperatures ( see sectin vii ) in the abscence of superconductivity , which is characteristic of a relativistic theory , and the mass @xmath0 can correspond to the pseudogap . as a model of htsc , the above kinetic term can also be motivated as follows@xcite .
suppose one is near the mott - hubbard insulating phase .
the anti - ferromagnetic phase of the heisenberg model has an effective low energy description in terms of a spin 3-vector field @xmath43 constrained to be of fixed length @xmath44 , with lagrangian @xmath45@xcite .
the field @xmath43 is bilinear in the fundamental electron fields , and in our model corresponds to @xmath46 .
the constraint on @xmath43 can be imposed by the constraint @xmath47 since [ 1.1b ] = - @xmath48 ^ 2 imposing these constraints one finds that [ 1.1c ] _ _ _ ^- _ ^+ + irrelevant operators which justifies the kinetic term in our model .
one can then relax the constraint on @xmath49 , and replacing it with a `` soft constraint '' by including a quartic interaction , as is done for the non - linear @xmath25 sigma model in two spatial dimensions .
another motivation for the kinetic term in our model ( at @xmath50 ) was given in @xcite based on the linear dispersion relation @xmath51 one obtains when expanding around a circular fermi surface .
however this involved a modification of , or at best a crude approximation to , the density of states .
canonical quantization and also a path - integral formulation follow from the euclidean action : [ 1.1 ] s = d^2 dt @xmath52 - 8 ^2 g ^-_^+_^-_^+ _ where @xmath53 .
the velocity @xmath22 plays the role of the speed of light , and it was proposed in @xcite that it be identified with the universal nodal fermi velocity@xcite . in the sequel we set @xmath54 except where indicated .
note that the fields @xmath10 are treated as lorentz scalars and spin is simply a flavor .
however it is possible to treat the fields as dirac spinors and thereby achieve complete lorentz invariance@xcite .
the quartic term is unique up to the sign of the coupling by fermionic statistics , and positive @xmath23 corresponds to repulsive interactions .
a consequence of the fermionic statistics is that the model has a hidden @xmath13 symmetry .
as explained in the introduction , the appearance of this @xmath13 is quite different from the @xmath13 symmetry proposed by zhang .
this symmetry is manifest if one considers an @xmath55-component version with fields @xmath38 , @xmath56 , which has @xmath57 symmetry ( hence the terminology `` symplectic fermions '' ) . for @xmath58 , since there are 4 fermionic fields and consequently a unique 4-fermion interaction , the theory automatically has @xmath59 symmetry .
the @xmath13 contains @xmath14 and @xmath15 subgroups which commute and can be identified with spin and electric charge respectively .
the conserved electric current then corresponds to [ current ] j^e_= -i _ @xmath60 and the fields @xmath61 have electric charge @xmath62 .
the important order parameters for the study of spontaneous symmetry breaking are composite bilinears in the fermions .
the 4 fields @xmath38 transform under the spinor representation of @xmath13 .
the bilinears can be decomposed as @xmath63 where @xmath64 is the singlet , @xmath18 the vector representation , and @xmath65 the adjoint .
the singlet is the field @xmath66 and corresponds to the mass term in the action .
the @xmath18-vector of fields corresponds to [ 1.2 ] = ( , ^+_e , ^-_e ) = ( ^- ^+ , ^+_^+ _ , ^-_^- _ ) where @xmath67 are pauli matrices .
the triplet of fields @xmath43 are electrically neutral and transform as a spin vector under the @xmath14 and serve as magnetic order parameters .
the fields @xmath68 on the other hand are spin singlets but carry electric charge @xmath19 and are thus cooper pair fields for superconducting order .
the @xmath13 invariant product is [ 1.3 ] = - 2 ^+_e ^-_e and the interaction can be expressed in the manifestly @xmath13 invariant manner : [ 1.4 ] _
int = g the momentum expansion of the free fields is [ momexp ] ^- ( , t ) & = & @xmath69 + ^+ ( , t ) & = & @xmath70 where @xmath71 and @xmath72 .
the canonical quantization of the theory based on the lagrangian leads to the canonical anti - commutations in momentum space : [ aadag ] \ { a _ , a^_ } = \ { b _ , b^_ } = ( 2 ) ^2 ( - ) and the hamiltonian is [ hamilton ] h_free = _ = , _ @xmath73 the @xmath74 and @xmath75 particles have opposite electric charge and can thus be thought of as particles and holes .
a distinguishing feature of our model , in contrast to quartic interactions of dirac fermions for instance , is that the quartic interaction is relevant in the renormalization group ( rg ) sense : the field @xmath10 has classical mass dimension @xmath11 so that the quartic interaction has dimension @xmath12 and the coupling @xmath23 dimension 1 .
this means that the interactions are important at low energies and can lead to non - fermi liquid behavior .
in fact , the model has a low energy rg fixed point , i.e. a quantum critical point at @xmath76 .
this critical point can formally be understood as an analytic continuation of the familiar wilson - fisher fixed point of the @xmath77 models , where @xmath78 , and some critical exponents , such as the anomalous dimension of the field @xmath10 , can be computed by specializing the known epsilon - expansion results for the @xmath77 model to @xmath78@xcite . however , many of the important operators such as the order parameters @xmath79 are composite fermion bilinears , which changes their structure compared to the magnetic order parameters of the @xmath77 models .
furthermore , @xmath79 includes bilinear order parameters of charge @xmath19 that have no counterpart in @xmath77 physics .
these rg properties are summarized in the next section . another interesting feature of our model is that whereas the fundamental interaction is repulsive , when one incorporates the momentum - dependent scattering at second order in perturbation theory ( 1-loop ) there is an instability toward the formation of a d - wave superconducting ground state .
this was studied in @xcite by deriving a d - wave gap equation for momentum - dependent vacuum expectation values for the cooper - pair fields @xmath68 .
this is reviewed briefly in section ix where we derive and study the finite temperature version of the d - wave gap equation .
the above free hamiltonian ( [ hamilton ] ) in momentum space is obviously hermitian and defines a unitary theory , in spite of the fact that the klein - gordon type of action normally associated with free bosons was quantized with the `` wrong '' statistics , i.e. as a fermion .
there are no negative norm states in the fock space of @xmath80 particles since there are no unwanted minus signs in eq .
( [ aadag ] ) . in @xmath81
the free massless symplectic fermion model is normally considered a non - unitary @xmath82 conformal field theory . the resolution of this apparent contradiction can be found in @xcite , where it was shown that the symplectic fermion can be mapped onto the @xmath83 unitary theory@xcite of dirac fermions at the level of detailed conformal partition functions . in light of the above results , the latter fact is not surprising since in momentum space the hamiltonians of symplectic and dirac fermions are identically eq . ( [ hamilton ] ) .
the issue of unitarity was further addressed in @xcite , where the main concern is the consistency of the interacting theory . because of the extra minus sign in ( [ momexp ] ) , the fields @xmath61 are not hermitian conjugates of each other , but rather [ unit.1 ] ^+ = c ( ^-)^c where the unitary operator @xmath84 simply distinguishes particles and holes : @xmath85 and @xmath86 , with @xmath87 .
the operator @xmath84 is easily constructed : [ ccons ] c = @xmath88 the interacting theory is thus pseudo - hermitian : @xmath89 .
the free theory is actually hermitian in momentum space since it is quadratic in @xmath75 s . as before there are no negative norm states in the fock space .
a pseudo - hermitian hamiltonian has real energy eigenvalues , and can still lead to a unitary time evolution , as discussed in @xcite .
let us make the following additional remarks concerning this issue in this specific context .
a unitary time evolution follows if one modifies the definition of hermitian conjugation .
define the @xmath84-hermitian conjugate as @xmath90 .
then the s - matrix is unitary with respect to this conjugation : @xmath91 since @xmath92 .
however the @xmath84-conjugate hermitian conjugation implies a modified definition of the inner product : @xmath93 . under this new inner product , states with an odd number of @xmath75 particles
now have negative norm . a completely equivalent description that dispenses altogether with the operator @xmath84 amounts to removing the minus sign in the expansion of @xmath94 in eq . ( [ momexp ] ) .
this gives a hermitian hamiltonian since now @xmath95 .
however this also gives the modified relation @xmath96 and this leads to the same conclusion that the @xmath75 particle sector contains negative norm states .
we now argue that this model can still yield a consistent theory at _
low energy_. let us use the description based on the @xmath84-operator ; the same arguments apply to the manifestly hermitian description without it .
the probability that an initial state @xmath97 evolves to a final state @xmath98 is given by [ prob ] p_if = states can be classified according to their @xmath84 eigenvalue of @xmath99 , where @xmath100 corresponds to negative norm .
one sees from the above formula that the probability @xmath101 is positive so long as the states @xmath102 are both of either positive or negative norm .
the problematic transitions with negative probability thus arise if the matrix element @xmath103 is non - zero for states of mixed norm .
however at low energies compared to the mass @xmath0 , these transitions are forbidden as a result of energy and charge conservation .
for example , suppose the initial state consists of only @xmath74-particles .
because of charge conservation , processes that change the number of @xmath75 particles necessarily involve creation of @xmath104 pairs ; for instance @xmath105 is not allowed .
the process @xmath106 is consistent with charge conservation , but not allowed kinematically , so one needs to consider at least @xmath107 .
energy conservation shows that the energy of the initial @xmath74 particles must be at least @xmath108 in order for the process to be kinematically allowed .
thus at low energies compared to the mass @xmath0 no @xmath75-particles will be produced from a state containing only @xmath74 particles .
a variation of this viewpoint interchanges the role of @xmath75 and @xmath109 , so that particles are created by the @xmath75 s and have negative energy .
one can now imagine that all the negative energy states are filled according to the pauli principle .
since there is a gap of @xmath108 between the negative and positive energy states , at low energies compared to @xmath0 no @xmath110 pairs will be created out of the dirac sea and one can deal only with the @xmath74-particles . as we describe in the sequel ,
the mass @xmath0 is on the order of the cut - off and can be quite large , typically greater than @xmath111 for htsc materials , so the low energy approximation is expected to be good for a large range of temperatures . at very high temperatures ,
our model necessarily breaks down and must be replaced by another effective theory that restores unitarity .
for instance , the cross - over to ordinary fermi liquid behavior in for example the specific heat is not built into the model . another way to obtain
a low energy unitary theory involves the introduction of a chemical potential , which is necessary at finite density .
particle number is not conserved here ; instead a chemical potential couples to the electric charge @xmath112 , i.e. @xmath113 .
the free hamiltonian is then [ hmu ] h = @xmath114 because of the difference in electric charge between the @xmath74 and @xmath75 particles , at a given @xmath115 there is a gap of @xmath116 between the energies of the @xmath74 s and @xmath75 s .
let us suppose that @xmath117 is chosen so that all of the @xmath75-states are filled . typically @xmath117 will be comparable to the fermi energy and typically very large . if @xmath117 is large enough , then because of the energy gap @xmath116 at low energies we can effectively deal only with the @xmath74-particle sector .
alternatively , one can view the @xmath74 sector as an in - accessible high energy sector when @xmath117 is large enough , and deal only with the low - energy @xmath75 sector . either way , as discussed above , all probabilities @xmath101 will be positive .
in all approaches to the renormalization group there are two cut - offs , a fixed upper cut - off @xmath21 typically related to the inverse lattice spacing , and a lower cut - off @xmath118 which is the running rg scale .
renormalization effectively removes the degrees of freedom with energy between @xmath119 and @xmath21 . in the high - energy physics context ,
the cut - off @xmath21 is unknown , so one normally performs the renormalization by sending @xmath21 to infinity and incorporating counter - terms to cancel ultra - violet divergences . in @xmath120 spacetime dimensions
this can be conveniently and systematically performed by using dimensional regularization in @xmath121 dimensions , which leads to the epsilon - expansion technique for studying models in @xmath122 . in the present context , since the cut - off @xmath21 is in principle known and physical quantities depend on it , it is more appropriate to perform the rg directly in @xmath122 with explicit cut - offs rather than use the epsilon expansion
. the dependence of the couplings on the running rg scale @xmath119 can be determined by considering an upper cut - off @xmath119 only .
consider first the coupling constant @xmath23 .
the 1-loop correction to the vertex is cancelled by a shift of @xmath23 , i.e. by defining @xmath123 where [ 2.2 ] g = 16 ^2 g^2 ^ when @xmath124 , @xmath125 . thus [ 2.3 ] _
g ( ) = let us define [ 2.5 ] g = then the beta function for the dimensionless coupling @xmath126 is [ 2.6 ] _
= - + 8 ^2 the above beta - function eq .
( [ 2.6 ] ) has a low energy fixed point at @xmath127 .
it is known that this fixed point survives , with small corrections , to higher orders@xcite .
we can now integrate the rg flow and incorporate the initial data at the high energy cut - off @xmath21 .
it will be convenient to introduce the dimensionless variable @xmath35 : [ 2.7 ] x 1/ the fixed point occurs at @xmath128 .
let @xmath129 be the value of the coupling at the scale @xmath21 , and @xmath130 .
then the solution to the rg flow equation ( [ 2.6 ] ) takes the simple linear form in @xmath35 : [ 2.8 ] = , ( x _ * = 8 ) there are two cases to consider depending on whether the coupling is strong @xmath131 or weak @xmath132 at short distances .
based on the relation between @xmath35 and hole doping described in section vi , we will refer to @xmath133 as the underdoped region and @xmath134 as overdoped ; in both cases @xmath135 .
furthermore , in order to clearly distinguish the two cases we will refer to @xmath136 in the overdoped region as @xmath137 .
( what we refer to as the overdoped region is close but not identical to the usual terminology ; the latter refers to the region beyond the maximum @xmath3 . )
it is clear that our model is characterized by a single fundamental energy scale set by the cutoff @xmath21 .
since @xmath21 has units of a wave - vector @xmath115 , this energy scale is [ energyscale ] e_0 = v_f _ c k_b t_0 where @xmath21 should be proportional to the inverse lattice spacing . before turning to mass renormalization , we now determine the 1-loop corrections to the anomalous dimension of the order parameters which are composite fields .
consider first the singlet operator @xmath138 . to first order in perturbation theory ,
[ 2.23 ] ^- ^+ ( 0 ) = 8 ^2 g @xmath139 ^- ^+ ( 0 ) this contribution is cancelled by renormalizing the operator as follows : [ 2.24 ] ^- ^+ z_^- ^+ where @xmath140 and [ 2.26 ] z _ = - 8 ^2 g ^ the anomalous dimension @xmath141 of @xmath142 is then [ 2.27 ] _
= _ z _ = - 4 where again we have assumed @xmath124 . repeating the above calculation for the @xmath13 vector of order parameters one finds _
= 4 let @xmath143 denote the scaling dimension of @xmath144 at the fixed point , including the classical contribution of @xmath11 for each @xmath10 field .
using @xmath145 at the fixed point , one obtains to 1-loop [ 2.29 ] , = = = thus one sees that the singlet is shifted down from the classical value , whereas the 5-vector of order parameters is shifted up .
the above values agree exactly with the 1-loop approximation in the epsilon - expansion obtained in @xcite .
we point out that although the actual values of the fixed point value @xmath146 and the functions @xmath147 differ in the calculation performed here compared to the epsilon expansion , the results for the anomalous dimensions at the critical point agree to 1-loop . from the two - loop results obtained in @xcite
, one sees that higher order corrections are reasonably small : @xmath148 changes to @xmath149 and @xmath150 actually remains at @xmath151 at the next order .
now we consider mass renormalization , which will be important in the sequel .
the 1-loop correction to the self - energy shown in figure [ figure2 ] is a negative correction to @xmath152 .
this is cancelled by @xmath153 where [ 2.20 ] m^2 = 8 ^2 g ^ the @xmath152 term on the rhs is extracted by taking a derivative with respect to @xmath152 and then setting @xmath154 : [ 2.21 ] m^2 = 4 g m^2 / defining [ mhat ] m = one finds _
^2 = ( -2 - 4 ) ^2 the @xmath155 term simply corresponds to classical dimension @xmath12 of @xmath152 and @xmath156 the quantum correction . since @xmath152 is the coupling for @xmath157 this implies that the anomalous dimension of @xmath142 is @xmath158 , in agreement with eq .
( [ 2.27 ] ) .
[ figure2 ]
although the mass @xmath0 is classically zero in the connection to the @xmath25 sigma model description of the heisenberg af described in section ii , the operator @xmath142 is relevant and not forbidden by any symmetries so that in general it will be generated when one includes interactions . in this section
we determine this dynamically generated mass to lowest order and propose that it be identified with the htsc pseudogap . if the original theory is massless , at 1-loop the correction to @xmath152 coming from the diagram in figure [ figure2 ] is negative and equal to @xmath159 .
since the propagator @xmath160 becomes @xmath161 when @xmath162 , a self - consistent equation for @xmath0 is the following [ 3.4 ] m^2 = - 8 ^2 g the one point function for the singlet is then [ 3.5 ] = -2 = where we have used the equation ( [ 3.4 ] ) .
the two above equations can also be derived from a hubbard - stratanovich transformation based on the fact that the interaction is proportional to @xmath163 , as was done in @xcite . in order to make sense of eq .
( [ 3.4 ] ) and obtain solutions , the mass renormalization discussed in section iv needs to be taken into account . taking the limits of integration to be from zero to @xmath119 , eq .
( [ 3.4 ] ) becomes [ 3.6 ] m^2 = -4 g + 4 g m ^-1 / m from eq .
( [ 2.20 ] ) with @xmath50 on the right hand side , one obtains @xmath164 , therefore the first term on the rhs above can be absorbed into @xmath152 , and this is consistent with renormalization .
the equation for @xmath0 now becomes [ 3.7 ] = 4 ^-1 1/ and has real solutions which are easily found numerically .
there are two useful analytic limits to the solution of eq .
( [ 3.7 ] ) . when @xmath126 is large , @xmath165 is also large and the solution is approximately : [ 3.8 ] 2 , ( large ) when @xmath126 is small , @xmath165 is also small and @xmath166 .
thus in this limit one has [ 3.9 ] 2 , g ( small ) it is interesting to note that if one sends the cut - off to infinity in eq .
( [ 3.4 ] ) and performs the integral by analytic continuation in the spacetime dimension , then one obtains the solution ( [ 3.9 ] ) .
this means that the analog of dimensional regularization here overestimates @xmath165 , since by eq .
( [ 3.7 ] ) , @xmath167 . in the underdoped region
the approximation ( [ 3.8 ] ) is considerably better than ( [ 3.9 ] ) and we will use it in places below .
the mass @xmath0 corresponds to an @xmath35-dependent energy scale [ epg ] e_pg ( x ) = v_f m(x ) k_b t^ * ( x ) where as before @xmath168 and @xmath169 is the solution to eq .
( [ 3.7 ] ) . in units of the fundamental scale @xmath20 : [ epg2 ] = = ( x ) = ( x ) in the next section we will relate @xmath35 to doping , thus the energy scale @xmath28 is doping dependent . it should be emphasized that @xmath28 simply corresponds to the temperature independent energy scale @xmath170 and is thus not a real temperature ; however as we will show in the sequel it can correspond to a cross - over scale in the real temperature .
a non - zero mass @xmath0 clearly corresponds to a gap in the density of states since the 1-particle energies are @xmath31 .
we discuss this further at the end of this section . as a model of htsc
we thus identify the mass @xmath0 with the pseudogap energy scale , i.e. @xmath171 . as we will show below , the thermodynamic properties of our model also support the identification of @xmath28 with the htsc pseudogap . a plot of @xmath28 verses @xmath35 is shown in figure [ figure3 ] .
note that @xmath28 is close to linear near the critical point at @xmath128 .
furthermore , the pseudogap is smaller on the overdoped side .
we point out that the re - appearance of the pseudogap on the overly doped side is contrary to what is normally observed . as a function of @xmath35 for @xmath172 and @xmath173.,width=377 ] [ figure3 ] since the overall energy scale @xmath20 depends both on @xmath22 and @xmath21 , in attempting to compare with the htsc data it is more useful to fix @xmath20 by using the pseudogap .
it is known experimentally that at zero doping in the region of the af phase @xmath170 is approximately the af exchange energy @xmath174 . since zero
doping occurs at @xmath175 ( see the next section ) , we can identify @xmath176 . since @xmath177 is of order 1 , @xmath178 .
for the cuprates @xmath179 .
an estimate of the density of states @xmath180 as a function of energy @xmath181 can be obtained as follows .
one has [ den.1 ] ( k ) dk = = = ( e ) de if we approximate @xmath182 in the above equation as the fermi wave - vector @xmath183 , then [ den.2 ] ( e ) @xmath184 ^ -1 = this behavior is qualitatively shown in figure [ figure4 ] . .
the dotted line represents a sketch of what is measured using arpes.,width=377 ] [ figure4 ] angle - resolved photoemission ( arpes ) measurements show a clear depression in the density of states and the leading edge of this depression is characterized by the pseudogap scale @xmath28 .
this behavior is shown schematically in figure [ figure4 ] as a dotted line .
thus one qualitative difference with what is observed is that the gap in our model is a clean one .
another difference is that due to the rotational invariance of our model , the pseudogap is isotropic , whereas the observed pseudogap shows lattice effects .
it is anisotropic in the direction relative to the fermi surface : the pseudogap is largest in the anti - nodal @xmath185 direction and vanishes along `` fermi arcs '' in the nodal @xmath186 direction .
since renormalization removes degrees of freedom between @xmath119 and @xmath21 , this suggests that doping can be varied by varying the cutoff @xmath119 . in @xcite
it was argued that a measure of hole doping can be defined based on the 1-point function @xmath187 .
since this one point function is proportional to @xmath119 by dimensional analysis , we are led to identify doping @xmath188 with the dimensionless quantity [ hole.1 ] p = @xmath189 for some constant @xmath190 . in the underdoped regime , the 1-point function is approximately given by eq .
( [ 3.8 ] ) : [ hole.2 ] p(x ) @xmath191 = @xmath192 where we have used eq .
( [ 2.8 ] ) .
half - filling then corresponds to @xmath193 , i.e. @xmath194 .
thus , in the approximation we have made , plots of various physical properties as a function of doping @xmath188 is simply related to plots as a function of the inverse coupling @xmath35 by rescaling and shift of the @xmath35-axis .
we choose to plot against @xmath35 since this more clearly reveals the rg properties .
we can give a rough estimate of the constant @xmath190 following an argument made in @xcite .
consider a lattice fermion model where @xmath195 is the local spin variable at lattice site @xmath196 , which is bilinear in the fermion operators .
one has [ hole.3 ] = - n_n_+ ( n_+ n _ ) where @xmath197 is the number of fermions of spin up or down at each site .
from this relation one sees that at half - filling @xmath198 , @xmath199 and @xmath200 , i.e. @xmath201 is constrained to be a spin @xmath37 vector and the model can be mapped onto the heisenberg model . dividing the above equation by the volume squared and taking the infinite volume limit , one finds that the right hand side is @xmath202 where @xmath203 are number densities . in our continuum model , @xmath201
is represented by the bilinear @xmath43 and one has the identity : [ hole.4 ] = - ( ^- ^+ ) ^2 comparing with the continuum limit of eq .
( [ hole.3 ] ) , one identifies @xmath204 , which corresponds to @xmath205 . at the critical point
this gives @xmath206 .
since this is only a rough estimate , one can alternatively fix @xmath190 in principle by fitting to experimental data .
an approximation to the specific heat can be made based on including just the effects of the dynamically generated mass .
a gas of 4 types of fermionic particles with single particle energies @xmath207 has the free energy per volume at temperature @xmath208 : [ sp.1 ] = -4 t _ 0^_c
@xmath209 one easily sees that @xmath210 is a function of the two dimensionless variables @xmath211 and @xmath212 where as in section v we have identified @xmath2 . at very low temperatures
@xmath213 we can approximate the @xmath214 as @xmath215 and effectively send the cut - off @xmath21 to infinity , and @xmath216 becomes a scaling function of @xmath217 .
the result is [ sp.2 ] - e^-t^*/t @xmath218 the entropy density is then : [ sp.3 ] s = - e^-t^*/t @xmath219 ^ 3 and the specific heat : [ sp.4 ] c = - t e^-t^*/t @xmath220 ^ 2 + @xmath221 ^ 3 it is convenient to define the quantity @xmath222 since for a fermi liquid @xmath223 is a constant .
the primary feature of our model is that @xmath224 at very low temperatures compared to @xmath28 and @xmath225 , which is ultimately attributed to the relativistic nature of the model . at higher temperatures
there is a crossover to a different behavior . since @xmath21 can be scaled out of the eq .
( [ sp.1 ] ) , the cross - over temperature is the pseudogap temperature @xmath28 . below we
plot the entropy and @xmath223 as a function @xmath208 for various `` doping '' @xmath35 .
the plots are in terms of the dimensionless quantities @xmath226 , @xmath227 and @xmath228 .
one clearly sees the crossover at @xmath36 . .
the vertical axis is the dimensionless quantity @xmath229.,width=377 ] [ figure5 ] as a function of temperature at various `` doping '' @xmath35 for @xmath172 .
the larger amplitude curves correspond to larger doping .
the vertical axis is the dimensionless quantity @xmath230.,width=377 ] [ figure6 ] experimental data for @xmath223 is shown in figure [ loramfig1 ] . in comparing with our results ,
it is important to bear in mind that the data contains contributions from the quasi - particle excitations in the sc phase , which explains the peaks to the left , and such effects are not included in our calculation .
thus , our curves should be compared with the data to the right of the sc peaks . at temperatures below @xmath225 one
sees a reasonably good qualitative agreement with the behavior computed above : for @xmath231 , @xmath232 , with a crossover to a different behavior at @xmath28 .
the dependence on doping is also qualitatively correct , i.e. the peaks of the curves move toward the left with increased doping .
one difference is that whereas our computed @xmath223 goes to zero at high temperature , the experiments indicate that at high temperatures @xmath223 approaches the fermi - liquid result , i.e. @xmath223 approaches a constant .
our model of course does not crossover to a fermi liquid at high temperatures , because as explained in section iii it is expected to break down at high enough temperatures , so at best it may describe temperatures up to and above @xmath28 . for @xmath233 ( from @xcite).,width=377 ] [ loramfig1 ]
in this section we compute the spin response to a magnetic field and also the magnetic field dependence of the specific heat in the same approximation we made in the last section , i.e. we only consider the effects of the dynamically generated mass .
both these quantities can be studied by adding a term to the action @xmath234 . comparing with the calculation of the effective potential @xmath235 in @xcite
, one sees that the free energy at zero temperature is [ mag.1 ] = - @xmath236 where as before , @xmath237 .
the one point function @xmath238 in the presence of @xmath239 can be obtained from the first derivative of the logarithm of the partition function @xmath240 where @xmath241 is the volume . for a single component of @xmath43 , one obtains [ mag.2 ] = = @xmath242 at finite temperature @xmath208
, @xmath243 becomes a quantized matsubara frequency @xmath244 where @xmath245 and one makes the replacement : [ mag.3 ] t _ z + 1/2 we need the identity [ mag.4 ] t _ z + 1/2 = @xmath246 this gives [ mag.5 ] = @xmath247 the linear response for small @xmath248 follows from taylor expanding the integrand : [ mag.6 ] = _ s ( m , t ) where [ mag.7 ] _ s ( m , t ) = @xmath249 plots of @xmath250 as a function of temperature are shown in figure [ figure7 ] .
one sees that the spin response is quenched when @xmath1 , consistent with experiments . for @xmath172 .
the vertical axis corresponds to the dimensionless quantity @xmath251.,width=377 ] [ figure7 ] define [ mag.8 ] = ( h ) - ( 0 ) integrating the identity eq .
( [ mag.4 ] ) one can show [ mag.9 ] t _ z + 1/2 @xmath252 = 2 t @xmath253 + - _ therefore up to a constant that is independent of @xmath208 and does not affect the specific heat : [ mag.10 ] = -2 t @xmath254 one can then define [ mag.11 ] = - plots of @xmath255 as a function of @xmath208 for different @xmath256 are shown in figures [ figure8 ] , [ figure8b ] .
( these plots are in arbitrary units since we have not specified the strength of the magnetic field . ) at very small @xmath248 , @xmath257
. evaluated at @xmath258 for @xmath259 as a function of temperature for fixed @xmath260 .
the larger amplitude curves correspond to larger @xmath35 , i.e. larger doping.,width=377 ] [ figure8 ] as a function of temperature for various @xmath261 ( increasing @xmath248 corresponds to increasing amplitudes on the plot ) at fixed @xmath262.,width=377 ] [ figure8b ]
it was shown in @xcite that the 1-loop corrections to scattering of cooper pairs leads to a d - wave superconducting gap of the form [ d.0 ] q ( ) = ^2 ( k_x^2 - k_y^2 ) = ^2 k^2 ( 2 ) where @xmath263 is a fourier transform of cooper pairing order parameters @xmath264 . before turning to the main topic
, some remarks on the d - wave property of the gap are called for . since the d - wave gap equation derived in @xcite
was based on a rotationally invariant hamiltonian , the most general solution involves a rotation by an arbitrary angle @xmath265 , i.e. @xmath266 in the above equation . on the other hand in htsc
the lattice breaks the rotational symmetry and the d - wave gap is oriented with respect to the fermi surface in a specific way , namely the gap vanishes in the nodal direction @xmath267 , which corresponds to @xmath268 .
this can be reconciled as follows .
free particles on the lattice have dispersion relation @xmath269 .
thus the fermi surface has the symmetry @xmath270 . requiring this symmetry fixes @xmath268 . in this sub - section
we study the effect of the non - zero mass @xmath0 on the zero temperature d - wave superconducting gap equation .
the effect of the mass term is simply the shift @xmath271 and the d - wave gap equation derived in @xcite becomes : [ d.1 ] ^4 = 2 g_2 ddk^2 @xmath272 where @xmath273 .
the coupling @xmath274 comes from the 1-loop scattering of cooper pairs and is given by [ d.2 ] g_2 = for simplicity we incorporate the cut - off as follows : @xmath275 and @xmath276 which is more appropriate for comparison with the finite temperature version we consider below .
re - expressing the gap equation in terms of dimensionless quantities by rescaling @xmath277 and @xmath278 , one obtains [ d.3 ] ^4 = 2 _ 2 _
0^d_0 ^ 1 dk^2 @xmath279 where @xmath280 and @xmath281 . since eq .
( [ d.2 ] ) is ultra - violet convergent , it can be approximated by letting the upper cut - off go to infinity , giving @xmath282 . incorporating the rg prescriptions of section iv , @xmath283 and @xmath284 , one finds that the parameters in the gap equation ( [ d.3 ] ) are the following : [ d.4 ] _ 2 = @xmath285 ^ 3 , m = where the ratio @xmath286 is given in eq .
( [ 2.8 ] ) . finally , since under a rg transformation @xmath287 , the physical gap in the theory at rg scale @xmath119 is @xmath288 .
the @xmath35-dependence of the solutions @xmath289 arises from the @xmath35-dependence of @xmath290 and @xmath291 .
a plot of @xmath292 as a function of x is shown in figure [ deltax ] .
it is important to point out that the interpretation of the critical point at @xmath293 presented here differs from the original proposal in @xcite , in that in the present work we extend @xmath35 beyond @xmath293 by introducing @xmath137 , and this places the maximum value of the gap near the critical point @xmath294 .
( we have effectively patched together what was referred to as type a and b in @xcite . ) further justification for this location of the critical point is based on the calculation of @xmath3 , which reaches a maximum value near the critical point , as explained below .
vs @xmath35 for @xmath295 and @xmath296.,width=377 ] [ deltax ] one can easily verify numerically that the pseudogap competes with sc in the following sense : if one artificially increases the mass @xmath0 , then the value of the gap @xmath289 decreases , and eventually vanishes for @xmath0 too large .
( this was shown already in @xcite . ) in order to derive the finite temperature version of the above d - wave gap equation , we start with the un - integrated form derived in @xcite : [ ft.1 ] q ( ) = - g ( , ) where the kernel is [ ft.2 ] g ( , ) = - 8 ^2 g_2 k^2 k^2 2(- ) the equation ( [ d.1 ] ) is obtained upon performing the angular integral . at finite temperature , @xmath243 becomes a quantized matsubara frequency @xmath297 , where @xmath208 is the temperature and @xmath298 is a half - integer , i.e. @xmath245 . as before , the integral @xmath299 is replaced with @xmath300 .
one needs the identity : [ ft.3 ] t _ = _ = @xmath301 where @xmath302 .
due to the specific form of the kernel @xmath303 , the solution to the equation ( [ ft.1 ] ) is of the d - wave form ( [ d.0 ] ) ( up to an arbitrary rotation ) where @xmath289 satisfies the integral equation [ ft.4 ] ^2 = g_2 dk d k^3 ( 2 ) @xmath304 where [ ft.5 ] _ , = finally , the physical temperature at the scale @xmath119 follows from the rg transformation @xmath305 .
solutions of the above equation are @xmath306 .
one can easily verify numerically that as the temperature goes to zero , one recovers the solution @xmath307 to the zero temperature gap equation ( [ d.3 ] ) .
one also finds that as the temperature is raised there are no solutions to the above equation for @xmath308 and this defines the @xmath35-dependent critical temperature @xmath3 .
this is shown in figure [ deltaoft ] . as a function of @xmath208 for @xmath309.,width=377 ] [ deltaoft ] the critical temperature @xmath3 as a function of @xmath35 on both sides of @xmath293
are shown in figure [ tc ] .
it turns out that @xmath3 at the critical value @xmath310 , i.e. @xmath311 , is universal in that it only depends on the overall scale @xmath225 and not on @xmath312 since the scale factor @xmath313 vanishes at this point . using the estimated relation between @xmath35 and doping in section vi , the critical point occurs at doping @xmath206 .
the dome shape of @xmath3 is a property of the mathematical structure of the gap equation and nothing universal is happening at the termination points of sc on either side of the critical point .
numerically we find that at the critical point [ ratio ] .084 , .268 we also find numerically that the maximum value of @xmath3 occurs close to the critical point so that @xmath314 , i.e. @xmath315 is simply proportional to fundamental energy scale @xmath20 .
as argued in section v , @xmath225 should be identified with the anti - ferromagnetic exchange energy @xmath174 at half - filling . for @xmath316 ,
this gives @xmath317 , which is quite reasonable for htsc .
it should be emphasized that the above @xmath3 is intrinsic to the two spatial dimensions , i.e. does not involve any kind of inter - planar energy scales . and @xmath318 vs @xmath35 for @xmath295 and @xmath296.,width=377 ] [ tc ]
in summary , we have further developed the interacting symplectic fermion model in two spatial dimensions by studying a dynamically generated relativistic mass and by including a finite temperature .
this allowed us to study some fundamental properties of the model , such as the specific heat and spin response , which clearly show non - fermi liquid properties .
as a simplified model of htsc , we identified the pseudogap energy scale with the zero temperature relativistic mass @xmath0 , and pointed out some close parallels with the observed phenomenology of the pseudogap .
we studied the effects of the pseudogap and finite temperature on the d - wave gap equation . in this model ,
the pseudogap clearly competes with superconductivity as a distinct phenomenon .
our analysis of @xmath3 suggests that the quantum critical point of our model , where the pseudogap vanishes , occurs inside the superconducting dome near optimal doping . for an antiferromagnetic exchange energy of @xmath319 , solutions of the d - wave gap equation
give a maximum @xmath3 of about @xmath5 .
we would like to thank ian affleck , denis bernard , dean robinson , henry tye , and jan zaanen for discussions .
this work is supported by the national science foundation under grant number nsf - phy-0757868 .
ek would also like to acknowledge the support of the national defense science and engineering graduate fellowship of the american society for engineering education .
j. w. loram , k. a. mirza , j. r. cooper and j. l. tallon , j. phys . chem .
solids vol * 59 * ( 1998 ) 2091 ; j. w. loram , j. luo , j. r. cooper , w. y. liang and j. l. tallon , j. phys .
solids vol * 62 * ( 2001 ) 59 ; j. l. luo , j. w. loram , t. xiang , j. r. cooper and j. l. tallon , arxiv : cond - mat/0112065 . | the interacting symplectic fermion model in two spatial dimensions is further analyzed . as an effective low energy theory ,
the model is unitary .
we show that a relativistic mass @xmath0 is dynamically generated and derive a gap equation for it . by incorporating a finite temperature we study some fundamental properties of the model , such as the specific heat and spin response , which clearly show non - fermi liquid properties .
we find that various physical properties are suppressed at temperatures @xmath1 where the cross - over scale is @xmath2 . as a simplified , toy model of high @xmath3 superconductivity ,
we thus identify the pseudogap energy scale with the zero temperature relativistic mass @xmath0 , and show that this reproduces some qualitative aspects of the observed phenomenology of the pseudogap .
the effects of the pseudogap and finite temperature on the d - wave gap equation are analyzed . in this model ,
the pseudogap is a distinct phenomenon from superconductivity and in fact competes with it .
our analysis of @xmath3 suggests that the quantum critical point of our model , where the pseudogap vanishes , occurs inside the superconducting dome near optimal doping . for an antiferromagnetic exchange energy of @xmath4 , solutions of the d - wave gap equation
give a maximum @xmath3 of about @xmath5 .
\def\ ] ] ] @xmath6 ) # 1#2#1 # 2 # 11 # 1 # 1#1 # 1#2#1 # 2 # 1#1 # 1 | # 1 pi # 1e^^#1 # 1_#1 # 1#2 # 1 , # 2 pi # 1e^^#1 # 1_#1 = cmss12 = cmu10 scaled1 # 1 pid(2)^d # 1|#1 # 1#1 | # 1#1 0=0=0 0 by-1exa |
the level of reported confidence intervals are most often 95% , with equal probability of missing the target at both sides . sometimes other levels are used , but rarely are several intervals at their different levels reported in applied work . instead of only reporting one confidence interval we suggest to report a family of nested confidence intervals for parameters of primary interest .
the family is indexed by the confidence level @xmath0 for @xmath1 and is conveniently represented by what is called a _ confidence curve _ , a quantity introduced by @xcite to give a complete picture of the estimation uncertainty . as an example , take @xmath2 for @xmath3 known .
it yields the curve @xmath4 for @xmath5 the cumulative distribution function of a @xmath6 .
this is a confidence curve since , for all @xmath1 , @xmath7 is the respective confidence interval of level @xmath0 . in the example
the confidence curve has its minimum at @xmath8 which is a point estimate of @xmath9 .
the normal confidence curve is tail - symmetric , i.e. the probability of missing the parameter to the left equals that to the right and is @xmath10 at level @xmath0 .
a tail - symmetric confidence curve represents uniquely a confidence distribution , that is confidence curves that describe upper confidence limits .
confidence distribution is a term coined by @xcite and formally defined in @xcite . for scalar parameters the fiducial distributions developed by @xcite are confidence distributions .
@xcite saw that the fiducial distribution leads to confidence intervals .
@xcite sees confidence distributions as `` simple and interpretable summaries of what can reasonably be learned from the data ( and an assumed model ) '' .
confidence distributions are reviewed by @xcite , and more broadly and with more emphasis on confidence curves by @xcite . in location models and other simple models
the confidence distribution is obtained from pivots , e.g. the normal pivot @xmath11 in the above example .
a canonical pivot is @xmath12 where @xmath13 is the distribution function of the maximum likelihood estimator @xmath14 , assumed to be absolutely continuous with respect to the lebesgue measure and non - increasing in @xmath9 .
see section [ section:2 ] for precise definitions and notation .
the confidence distribution @xmath15 is a canonical pivot in the sense of being uniformly distributed on the unit interval when @xmath14 is distributed according to @xmath9 .
when @xmath16 is a sufficient statistic with monotone likelihood ratio , @xmath15 is also optimal in the neyman - pearson sense , that is it describes smaller confidence intervals at a given level when compared to any other confidence distribution for the parameter @xmath9 ( * ? ? ?
* section 5.4 ) .
an equal tailedconfidence curve is readily obtained from @xmath15 by @xmath17 . in this paper
we shall be concerned with confidence curves obtained from the log - likelihood ratio @xmath18 , and we shall study the properties of median bias correction .
median bias correction of a confidence curve , proposed by @xcite , is a method to make the resulting confidence curve approximately tail symmetric . in the normal example @xmath19 and
the confidence curve mentioned above is also given by @xmath20 where @xmath21 is the cumulative chi - square distribution function with one degree of freedom .
this confidence curve is tail - symmetric , as mentioned , and the confidence interval of level @xmath22 is the single point @xmath16 which thus has median @xmath9 and is said to be median unbiased .
in general @xmath18 hits zero at the maximum likelihood estimator , which might not be median unbiased .
let @xmath16 have median @xmath23 .
the median bias corrected confidence curve is the confidence curve of the parameter @xmath23 .
the idea is to probability transform the bias corrected log likelihood ratio @xmath24 rather than @xmath18 . with @xmath25 denoting the sampling distribution of @xmath26 when the data is distributed according to @xmath9 , the bias corrected confidence curve is @xmath27 . since @xmath28 is median unbiased , the level set at @xmath29 is typically the single point @xmath30 and , by continuity , @xmath31 is close to be equal tailedat low levels .
we undertake a theoretical study of the asymptotic properties of @xmath31 by showing that @xmath31 is third - order tail - symmetric for @xmath32 in the normal deviation range in two important classes of parametric models with parameter dimension one .
first , we consider parametric models that belong to the efron s normal transformation family @xcite .
then , we extend the result to regular one dimensional exponential families , where we also discuss the relation between median bias corrected and modified directed likelihood of @xcite , thus providing an alternative approximation to the latter . since median bias correction works so well in these cases , it is reasonable to expect the method to work well quite generally .
however , when a canonical confidence distribution is available , as in the exponential family models , we do of course not advocate to use median bias correction rather than using the canonical confidence distribution .
the rest of the paper is organized as follows . in section [ section:2 ]
, we recast confidence estimation in terms of confidence curves and introduce the notation we use in the sequel .
we also define the confidence curve based on inverting the median bias corrected version of the log - likelihood ratio . in section [ section:3 ] and [ section:4 ]
, we investigate its asymptotic properties in terms of tail symmetry in the efron s normal transformation family and in one dimensional exponential families , respectively .
finally , in section [ section:5 ] some concluding remarks and lines of future research are presented , together with an example that provides a preliminary illustration of the use of median bias correction in the presence of nuisance parameters . some proofs and a technical lemma
are deferred to the appendix .
let @xmath33 be a continuous random sample with density @xmath34 depending on a real parameter @xmath35 and let @xmath36 indicate probabilities calculated under @xmath34 .
the log - likelihood is @xmath37 , and the log - likelihood ratio is @xmath38 , where @xmath16 is the maximum likelihood estimate .
we drop the second argument in sample - dependent functions like @xmath39 and @xmath40 whenever it is clear from the context whether we refer to a random quantity or to its observed value . unless otherwise specified , all asymptotic approximations are for @xmath41 and stochastic term @xmath42 refers to convergence in probability with respect to @xmath34 .
we assume that the model is sufficiently regular for the validity of first order asymptotic theory , cfr .
* chapter 3 ) .
in particular , @xmath18 converges in distribution to a chi - squared random variable , hence , by contouring @xmath18 with respect to this distribution we obtain intervals of @xmath9 values given by the level sets for the curve @xmath43 where @xmath21 is the distribution function of the chi - squared distribution with 1 degree of freedom .
this curve depends on the sample @xmath44 and has its minimum at @xmath16 .
however its level sets are not in general exact confidence intervals since the chi - squared approximation for the distribution of @xmath18 is valid only for @xmath45 large and the coverage probabilities equal the nominal levels only in the limit . as a consequence ,
@xmath43 is not uniformly distributed on the unit interval under @xmath46 , a property we require for a _
regular _ confidence curve as spelled in the following definition .
[ definition:1 ] a function @xmath47 is a regular confidence curve when @xmath48 , the level sets @xmath49 are finite intervals for all @xmath50 , and @xmath51 under @xmath46 .
confidence curves might be defined for parameters of higher dimension and also for irregular curves that even might have more than one local minimum or might have infinite level sets for @xmath52 , see ( * ? ? ?
* section 4.6 ) .
note that , under definition [ definition:1 ] , @xmath53 is an exact confidence region of level @xmath0 since @xmath54 . among confidence curves , of special importance
are confidence distributions , which are confidence curves that describe upper confidence limits .
the definition is as follows .
[ definition:1b ] a function @xmath55 is a confidence distribution when @xmath56 is a cumulative distribution function in @xmath9 for all @xmath44 and @xmath57 under @xmath46 .
keep in mind that the realized confidence curve and confidence distribution depend on the data , and prior to observation they are random variables ( with distribution depending on the parameter value from which the data are generated ) . to keep the notation simple
, we drop the second argument @xmath44 in @xmath58 and @xmath59 .
moreover , we will confine ourselves to regular confidence curves @xmath60 with only one local minimum . in this setting @xmath60
can be transformed into a distribution via @xmath61 so that the left and right endpoints of the interval @xmath62 , are given by @xmath63 respectively .
we refer to @xmath64 as the median confidence estimator for @xmath9 . by construction , @xmath65 and @xmath66
satisfy @xmath67 .
we then say that @xmath60 is tail
symmetric when the interval @xmath68 is equal tailed , that is @xmath69 this is equivalent to @xmath70 defining a confidence distribution according to definition [ definition:1b ] .
[ definition:2 ] a confidence curve @xmath60 is tail
symmetric if @xmath70 in is a confidence distribution according to definition [ definition:1b ] .
the relation obviously works in the other direction : given a confidence distribution @xmath15 , @xmath71 defines a tail - symmetric confidence curve , see @xcite .
note that the median confidence estimator @xmath72 of a tail
symmetric @xmath60 is median - unbiased , i.e. @xmath73
. see ( * ? ? ?
* section 5.6 ) .
the relation between median - unbiased estimators and equal tailedintervals have been noted by @xcite in connection with the maximum likelihood estimator .
we now focus on confidence distributions derived from the likelihood .
it is convenient to set as the exact confidence distribution the one obtained from the sampling distribution of the maximum likelihood estimator , namely @xmath74 where we assume that @xmath75 , the distribution function of @xmath16 , is continuous and non - increasing in @xmath9 . in order to have @xmath15 being a proper cumulative distribution function
, it is also required that @xmath76 and @xmath77 , where @xmath78 and @xmath79 are the infimum and supremum of the parameter space @xmath80 , respectively .
the @xmath81-quantile is denoted by @xmath82 .
in particular , @xmath83 corresponds to the median unbiased estimator of @xmath9 .
the exact distribution @xmath84 is generally unknown and the asymptotic approximation of the confidence limit @xmath85 has been object of an extensive research which goes beyond first order accuracy .
see @xcite for a review .
third order approximations to @xmath84 , and thus to @xmath15 , can be obtained from the modified directed likelihood of @xcite , see section 4.2 for a discussion .
we will instead look for a route to such good approximations by transforming the scale at which the log - likelihood ratio is presented . to this aim ,
let @xmath86 be the sampling distribution function of @xmath18 under @xmath46 , and define @xmath87 according to , @xmath65 and @xmath66 are the endpoints of a confidence interval of level @xmath0 .
it is clear that , in general , @xmath60 is not tail symmetric according to definition [ definition:2 ] , in particular when @xmath16 is not median unbiased .
more generally , the distribution estimator @xmath88 is not uniformly distributed on the unit interval under @xmath46 . according to first - order asymptotics
, @xmath60 is tail
symmetric up to the first order of approximation , that is @xmath89 consequently @xmath90 and @xmath91 and @xmath92 . in order to improve on ( [ eq : first - asymp2 ] ) , we consider the median bias correction to @xmath18 .
let @xmath23 be the median of @xmath16 as function of @xmath9 , that is @xmath93 by assumption , @xmath23 is continuously increasing in @xmath9 and @xmath94 , as a simple calculation reveals .
the _ median bias corrected log - likelihood ratio _ is defined as @xmath95 by construction , @xmath26 attains its minimum at @xmath83 , the median unbiased estimator of @xmath9 . since both the likelihood function @xmath96 and the median function @xmath23 are invariant to monotone parameter transformations , invariance is preserved for @xmath26 .
see @xcite for a different type of likelihood correction , aimed at reducing the bias of the maximum likelihood estimator .
the median bias corrected confidence curve is defined as @xmath97 where @xmath25 stands for the sampling distribution of @xmath26 under @xmath46 . according to
, it yields the distribution estimator @xmath98 for illustration , we consider confidence distributions for the variance parameter @xmath9 in the normal model @xmath99 . for @xmath100 ,
the log - likelihood ratio is @xmath101 .
based on @xmath102 , one finds @xmath103 and @xmath104 ( in obvious notation ) . using the pivotal distribution of @xmath105 , @xmath86 and @xmath25
can be computed via monte carlo .
based on a simulated sample of size @xmath106 with @xmath107 , the left panel of figure [ fig3 ] displays @xmath31 according to while the right panel reports @xmath15 , @xmath70 and @xmath108 , according to , and , respectively . in the normal model @xmath99 for @xmath106 observations generated according to @xmath107 .
left panel : @xmath31 together with some of its confidence intervals .
right panel : @xmath70 ( dashed line ) , @xmath108 ( solid line ) , and nearly on top @xmath15 ( dotted line ) .
@xmath31 , @xmath70 and @xmath108 are based on 50000 monte carlo simulations . _
_ ] note that @xmath108 and @xmath15 are on top of each other and are almost indistinguishable .
hence , the median correction in , by making the median confidence estimator of @xmath31 coincide with @xmath83 , shifts the whole curve @xmath108 towards @xmath15 , thus inducing nearly exact tail symmetry .
we return to this example in section [ section:4 ] where we give a theoretical justification to the fact @xmath108 and @xmath15 coincide to the third order of approximation in theorem [ theorem:2 ] .
we conclude this section by noting that , while @xmath109 can be interpreted as the log - likelihood ratio for the parameter @xmath110 , that is @xmath111 , @xmath112 does not correspond to the confidence curve in the @xmath113-parametrization , that is @xmath114 where @xmath115 stands now for the sampling distribution of the log - likelihood ratio in terms of the @xmath113 parameter . as an example , consider the exponential model , @xmath116 . by standard calculation
one finds that @xmath117 , @xmath118 and @xmath119 .
hence , for @xmath120 , we get @xmath121 while @xmath122 , where @xmath123 and @xmath124 . on the other hand , @xmath31 shares with @xmath60 the property of invariance with respect to monotone transformation of the parameter : if @xmath125 for @xmath126 invertible , then , it is easy to see that bias corrected confidence curve in the @xmath113parametrization , say @xmath127 , corresponds to @xmath128 .
this can be easily verified in the exponential model above by taking , e.g. , @xmath129 so that @xmath113 represents the mean parameter . in the sequel , for ease of notation
, we avoid superscripts as in @xmath130 and @xmath131 whenever the parametrization the likelihood is referring to will be clear from the context .
in this section we establish third order tail symmetry of the bias corrected confidence curve @xmath31 when @xmath16 is a sufficient statistic and belongs to the normal transformation family of @xcite .
this family of distributions was used by @xcite to introduce bias and acceleration corrected bootstrapped confidence intervals that achieve second order accuracy .
the idea is that standard intervals are based on assuming that the normal approximation of @xmath132 is exact , with @xmath133 a fixed constant and , hence , convergence to normality can be improved by considering a monotone transformation of @xmath16 and @xmath9 which is exactly normalizing and variance stabilizing .
second order accuracy was later extended to regular statistical models such as the exponential family , see @xcite .
we follow a similar path here , as we first prove , in theorem [ theorem:1 ] , tail symmetry in the normal transformation family as this case provides a simple illustration of the generalized inverse mapping argument reported in lemma [ lemma:1 ] of the appendix .
theorem [ theorem:2 ] of section [ section:4 ] addresses tail symmetry in the exponential family , where an additional cornish fisher expansion of the distribution of the maximum likelihood estimator is needed .
theorem [ theorem:2 ] is indeed a more general result than theorem [ theorem:1 ] since , by pitman - koopman - darmois theorem , cfr .
* theorem 6.18 ) , if the data are independent and identically distributed and the dimension of the sufficient statistic does not depend on @xmath45 , as we are assuming here , then the model is an exponential family . let @xmath16 be a sufficient estimator for @xmath9 , not necessarily maximizing the likelihood , but behaving asymptotically like the maximum likelihood estimator in terms of order of magnitude of its bias , standard deviation , skewness , and kurtosis : @xmath134 where @xmath135 , and @xmath136 are functions of @xmath9 and @xmath45 ( the latter suppressed in the notation ) bounded in @xmath45 . see equations ( 5.1)(5.3 ) in @xcite .
next , suppose there exists a monotone increasing transformation @xmath126 and constants @xmath137 ( bias constant ) and @xmath78 ( acceleration constant ) such that @xmath138 and @xmath139 satisfy @xmath140 where @xmath141 when @xmath142 and @xmath143 when @xmath144 .
model has standard deviation linear in @xmath145 on the transformed scale .
it provides a pivot with accompanying confidence distribution @xmath146 .
the latter is directly transformed back to a confidence distribution for @xmath9 , that is @xmath147 .
theorem [ theorem:1 ] states that @xmath148 as well as @xmath149 , are third order tail - symmetric according to definition [ definition:2 ] , an improvement up to @xmath150 in the asymptotic order displayed in .
the proof relies on the asymptotic inversion of convex functions reported in lemma [ lemma:1 ] in the appendix .
[ theorem:1 ] let @xmath16 be a sufficient estimator of @xmath9 based on a sample of size @xmath45 satisfying , and assume there exists a monotone increasing function @xmath126 such that holds .
then , for @xmath15 and @xmath108 defined in and , respectively , @xmath151 since a confidence curve for @xmath152 translates into one for @xmath153 for the invertible transformation @xmath126 , it is sufficient to prove in the transformed normal model . under , the normalizing transformation @xmath126 is locally linear in its argument with a scale factor of order @xmath154 . in particular , from , the normal deviation range @xmath155 in corresponds to @xmath156 according to ( * ? ? ?
* theorem 2 ) , @xmath157 $ ] and both @xmath158 and @xmath78 are @xmath159 as long as the @xmath16 satisfies .
we will make repeatedly use of these asymptotic behaviors throughout the proof , even though we suppress the dependence of @xmath78 and @xmath160 on @xmath45 in the notation .
the log - likelihood @xmath161 ^ 2/2-\log(1+a\phi)$ ] is not maximized at @xmath162 , unless @xmath163 , rather at @xmath164\ ] ] as a simple calculation reveals .
one finds that @xmath165 and , consequently , @xmath166 .
actually , @xmath167 belongs to the normal transformation family since it can be written as @xmath168 for @xmath169 , see ( * ? ? ?
* section 11 ) , with distribution @xmath170 and median function @xmath171 .
note that @xmath172 is increasing in @xmath173 when @xmath174 , which we assume without loss of generality since it certainly is for large @xmath45 .
since @xmath167 is a sufficient statistic , the log - likelihood ratio for @xmath173 is @xmath175 it is easy to check that @xmath176 is convex in both arguments , and so is its bias corrected version @xmath177 .
let @xmath178 be defined according to .
we are interested in expressing @xmath178 in terms of tail probabilities associated to @xmath167 for comparison with the confidence distribution @xmath179 . to this aim ,
let @xmath180 be implicitly defined in function of @xmath167 and @xmath173 by @xmath181 .
then , @xmath182 for @xmath183 when @xmath184 , for @xmath185 when @xmath186 .
we only consider the first case , where the equality of interest is @xmath187 hence , for @xmath184 , the normal deviation range @xmath155 in corresponds to @xmath188 for @xmath189 in . as for the right hand side of , from
it follows that , when @xmath184 , @xmath190 , so that is implied by @xmath191 for @xmath189 in . in order to establish
, we derive an asymptotic expansion of @xmath192 locally at @xmath193 by an application of the generalized inverse mapping argument of lemma [ lemma:1 ] .
let @xmath194x\big)$ ] so that @xmath195 for @xmath196/[1+ab(\phi)]$ ] . also , let @xmath197 be implicitly defined by @xmath198 so that @xmath199/[1+ab(\phi)]$ ] .
one finds @xmath200 $ ] and @xmath201 for @xmath202 , so that the hypotheses of lemma [ lemma:1 ] are satisfied .
hence , @xmath203 for @xmath204 , that is @xmath205 for @xmath206/[1+ab(\phi)]=o_p(1)$ ] . since @xmath207(1+a\phi ) = ( 1-az_0-c)(1+a\phi)$ ] and both @xmath208 and @xmath209 are @xmath210 , we get for @xmath206/(1+a\phi)=o_p(1)$ ] .
the latter corresponds to @xmath189 in the normal deviation range upon substitution for @xmath167 and @xmath172 .
the proof is then complete . as an illustration of ,
assume the coefficients in are in agreement so that @xmath211 is the maximum likelihood estimator , that is @xmath163 , see equation
. therefore , let @xmath212 , which yields @xmath213 .
the distribution function of @xmath211 is @xmath214 $ ] , so that @xmath215 $ ] , with median unbiased estimator @xmath216 . the log - likelihood ratio and its bias corrected version @xmath217 are easily derived .
the exact distribution @xmath218 of @xmath109 is recovered from the standard normal distribution , @xmath219 ^ 2 -2\log[1+az/(1-a^2)]$ ] , and @xmath220 can be calculated by monte - carlo . in the left panel of figure [ normal_transformation ]
we plot @xmath221 and @xmath178 for @xmath222 and @xmath223 . even for a non -
negligible acceleration @xmath78 ( later we argue that @xmath224 , so it roughly corresponds to @xmath106 ) , the median corrected confidence curve @xmath220 nearly exactly recovers , through @xmath178 , the confidence distribution @xmath221 .
the right panel shows that the difference between the two confidence distributions is very small , approximately of order @xmath225 , suggesting that the order of magnitude in might be conservative . , @xmath222 .
left panel : confidence distributions @xmath178 ( solid line ) , and nearly on top @xmath221 ( dotted line ) .
right panel : difference @xmath226 .
@xmath227 is based on 100000 monte carlo simulations . _ ]
_ _ 4.1 tail symmetry.__@xmath228 in this section we establish third order tail symmetry for the mean value parameter of regular one - parameter exponential families . following ( * ? ? ?
* section 5 ) , let @xmath229 $ ] , where @xmath230 is an unknown real parameter . given a random sample of size @xmath45 , the log - likelihood for @xmath230 based on @xmath231 has form @xmath232 $ ] . upon defining @xmath233 and @xmath234 ,
the log - likelihood for @xmath235 is @xmath236 .
since the cumulant generating function for @xmath237 is @xmath238 the @xmath239-th order cumulant of @xmath237 is @xmath240 , the @xmath239-th order derivative of @xmath241 .
we set @xmath242 , so that @xmath243 and @xmath244
. consequently , @xmath245 is the standard error of @xmath16 , where we use the subscript @xmath9 in @xmath246 to highlight the dependence on @xmath9 .
note that @xmath247 since @xmath248 .
the following result can be stated .
[ theorem:2 ] let @xmath16 and @xmath18 be the maximum likelihood estimator and the log - likelihood ratio for the mean value parameter in a continuous one - dimensional exponential model based on a random sample of size @xmath45 .
also , let @xmath246 be the standard error of @xmath16 and @xmath15 and @xmath108 be defined in and , respectively .
then , as @xmath41 , @xmath249 the proof is deferred to the appendix and we only provide here in this paragraph a sketch .
reasoning as in the proof of theorem [ theorem:1 ] , take @xmath250 so that @xmath251/2 $ ] , where @xmath252 is implicitly defined by @xmath253 .
when for @xmath250 , corresponds to @xmath254 , cfr . .
an asymptotic expansion of @xmath252 as function of @xmath16 and @xmath9 is obtained via the generalized inverse mapping result of lemma [ lemma:1 ] . in order to work with left tail probabilities ,
we further define @xmath255 as function of @xmath16 and @xmath9 to satisfy @xmath256 . while for the normal transformation family @xmath255 can be derived in explicit form , in the present setting an additional asymptotic expansion is required .
this is achieved by an edgeworth expansion and the allied cornish fisher inversion for the standardized distribution of @xmath257 .
the proof is completed by showing that @xmath252 and @xmath255 coincide up to the required asymptotic order for @xmath258 . note that tail symmetry of @xmath31 in the @xmath259 example of section [ section:2 ] holds by theorem [ theorem:2 ] for @xmath260 , and , hence , a theoretical justification of what we observed in figure [ fig3 ] is obtained .
it is worth noting that the chi - squared distribution of the maximum likelihood estimator is the running example in @xcite where it is shown that the transformation @xmath126 leading to nearly exists , see remark e in section 11 therein ( actually , @xcite considers the sampling distribution of the unbiased estimate of the variance when the mean is unknown ) .
hence , this example also provides an illustration of tail symmetry in the normal transformation family as stated in theorem [ theorem:1 ] . _ _ 4.2 comparison with the modified directed likelihood.__@xmath228 we adopt here the notation in ( * ? ? ?
* section 5 ) for the partial derivatives of @xmath40 with respect to @xmath9 and @xmath16 , namely @xmath261 for nonnegative integers @xmath262 and @xmath263 .
we also adopt the convention of a slash through @xmath40 indicating the substitution of @xmath9 for @xmath16 and a hat sign indicating the substitution of @xmath16 for @xmath9 after any differentiation .
the observed information is defined either as @xmath264 or as @xmath265 according to whether it is considered as a quantity depending on the parameter or the data only .
recall the definition @xmath266 the log - likelihood ratio and @xmath267 for the directed likelihood . since @xmath268 is increasing in @xmath16 , @xmath269 , where @xmath15 has been defined as @xmath270 assuming that the mle @xmath14 has distribution @xmath13 non - increasing in @xmath9 .
the modified directed likelihood is defined as @xmath271 see ( * ? ? ?
* section 6.6 ) .
it is a higher order pivot , that is it has normal distribution with error @xmath272 in the normal deviation range @xmath273 , so that @xmath274 .
consider now the median bias corrected directed likelihood , @xmath275 where we recall that @xmath23 has been defined as the median of the @xmath16 , i.e. the function of @xmath9 that satisfies @xmath276 .
notice that , since @xmath277 is increasing in @xmath16 , we also have that @xmath278 . in the next theorem we establish that , in regular one parameter exponential families , @xmath277 and the modified directed likelihood @xmath279 are second order equivalent in the normal deviation range . [ theorem:3 ]
let @xmath16 and @xmath18 be the maximum likelihood estimator and the log - likelihood ratio for the mean value parameter in a continuous one - dimensional exponential model based on a random sample of size @xmath45 .
also , let @xmath246 be the standard error of @xmath16 and @xmath279 and @xmath277 be defined according to . then , as @xmath41 , @xmath280 the proof is deferred to the appendix .
note that , because of the higher order pivotal property of @xmath279 , implies that @xmath281 in the normal deviation range , that is @xmath277 has sampling distribution closer to normality than @xmath268 .
there has been a renewed interest in confidence distributions in recent years , see @xcite and @xcite . in this paper
we have undertaken an asymptotic investigation of the merits of median bias correction in deriving higher order accurate confidence limits .
we found that , in regular one - dimensional models , the confidence distribution obtained from the bias corrected log - likelihood ratio is third order equivalent to the unique exact confidence distribution based on the maximum likelihood estimator .
moreover , the bias corrected directed likelihood provides a second order approximation to the modified directed likelihood of @xcite , thus consisting in a high order pivot .
it shows , from a different perspective , the close connection between the log - likelihood ratio and the distribution of the maximum likelihood estimator so extensively studied in the literature , a key example being the @xmath282 approximation of @xcite .
we are not aware of similar results in the literature on higher order asymptotics .
we have not discussed the effect of the bias correction on the sampling distribution of the log - likelihood ratio @xmath283 . with the median function @xmath23 at hand
, the chi - squared transformation of @xmath109 will typically provide more equal tailedintervals than the usual chi - squared calibration of @xmath39 .
we found however that median bias correction is second order equivalent to what is found via the modified directed likelihood .
the convergence to chi - squared distribution of @xmath109 is thus at least to the second order in regular one - parameter exponential families .
an important direction for future research is the extension of the results of theorem [ theorem:2 ] to models with nuisance parameters . in full @xmath284-dimensional exponential models
when the interest parameter @xmath9 is a linear function of the canonical parameters , or a ratio of two canonical parameters , a reparametrization from the canonical parameter vector @xmath235 to @xmath285 , where @xmath286 is a @xmath287-dimensional nuisance parameter , can be made and the canonical statistic @xmath237 can be re - expressed as @xmath288 having density @xmath289 $ ] .
exact inference on @xmath9 can be based on the conditional distribution of @xmath290 given @xmath291 , which depends on @xmath235 only through @xmath9 .
see @xcite , and @xcite who find the conditional confidence distribution to be uniformly most powerful .
the definition of @xmath15 and @xmath23 are to be interpreted conditionally on @xmath291 as well .
we expect the median bias corrected confidence curve based on the profile likelihood to be tail - symmetric to the third order , and to the second order to be chi - square distributed .
the investigation of the relation of the bias corrected profile likelihood with other versions of adjusted profile likelihoods that have been proposed in the literature would also be of interest . outside the exponential family ,
the evaluation of sample space derivatives of the likelihood requires the identification of an ancillary statistic .
moreover , the distribution of the maximum likelihood estimator has to be evaluated conditionally upon this statistic .
the asymptotic approximations used in theorem [ theorem:2 ] can be adapted to this setting , a natural extension being for transformation families .
next is a preliminary illustration of the use of median bias correction to confidence curves in a multidimensional statistical model .
the model in the example below is not in the exponential family , nor an ancillary statistic is available , and we there use brute force to handle the nuisance parameter . _
_ example.__@xmath228 we consider the `` bolt from heaven '' data example from section 7.4 in @xcite .
data consists of @xmath292 winning times in the fastest @xmath293-m races from 2000 to 2007 , that is races that clocked at @xmath294 seconds or better .
@xcite translate these races results @xmath295 as @xmath296 in order to apply extreme value statistics . specifically , the data is modeled using the generalized pareto distribution ( gpd ) which has density @xmath297 for @xmath298 .
sections 3.4 and 6.5 in @xcite .
interest is in estimating @xmath299 for @xmath300 and @xmath301 .
it takes on the interpretation of the probability , as seen at the start of 2008 , that in the @xmath302 fastest races of 2008 one should experience a race of @xmath303 or better , where @xmath303 is the world record time scored by usain bolt on 31 may 2008 .
see @xcite for details .
the authors compute a confidence curve for the parameter @xmath284 by profiling the log - likelihood , @xmath304 and by inverting the profile log - likelihood ratio @xmath305 with respect to the chi - squared distribution after bartlett correction , @xmath306 where @xmath307 ( found through simulations ) and @xmath308 is the chi - squared distribution function with @xmath309 degree of freedom . by construction
, @xmath310 points at @xmath311 according to maximum likelihood estimates @xmath312 @xmath313 and @xmath314 @xmath315 ( with approximate standard errors in parentheses ) and has 90% confidence interval @xmath316 $ ] .
we proceed next with median bias correction of @xmath317 so to produce the bias corrected confidence curve @xmath318 . to this aim
, the median function of @xmath319 needs to be estimated .
the problem here is that , since @xmath319 is not a sufficient statistics for @xmath284 , its sampling distribution is not uniquely determined by the value of @xmath284 .
if an ancillary statistic for @xmath284 was available , say @xmath320 , then the median function to be used would be @xmath321 where conditioning is intended with respect to the realized value of @xmath263 in the data .
note that @xmath322 does not depend on the parameter @xmath78 since the conditional distribution of @xmath319 given @xmath263 does not depend on @xmath78 by definition
. the median bias corrected log - likelihood ratio would then be @xmath323 with sampling distribution @xmath324 so that the bias corrected confidence curve would be @xmath325 .
for the sake of illustration we proceed here without relying on such ancillary statistic .
it is convenient to introduce the reparameterization @xmath326 so to define the median function of @xmath319 to depend on both @xmath284 and @xmath78 , say @xmath327 in the left panel of figure [ fig : bolt ] we plot @xmath328 for @xmath329 ( this range has about @xmath330 confidence for @xmath78 ) .
each curve is obtained by spline interpolation ( constrained to be @xmath22 at @xmath331 ) of the median of @xmath319 for a fine grid of @xmath284 values . for this we used monte carlo simulations : for each combination of @xmath78 and @xmath284 , 15000 samples of size @xmath292 from the gpd were drawn and @xmath328 is estimated via the 15000 realizations of the sample median .
hence we replace @xmath322 with @xmath332 , i.e. the solid line plotted in the left panel of figure [ fig : bolt ] .
the median bias corrected log - likelihood ratio is then defined as @xmath333 and we estimate its sampling distribution @xmath334 for a grid of @xmath284 values through simulations .
finally , we compute the bias corrected confidence curve @xmath335 and we plot it together with @xmath310 ( based on the chi - squared approximation with bartlett correction ) in the right panel of figure [ fig : bolt ] . median bias correction moves the confidence curves slightly to the right to the effect that the upper 5% confidence quantile is @xmath336 instead of @xmath337 .
@xmath338{median.pdf } & \includegraphics[width=0.45\textwidth]{fig7_3new.pdf } \end{array}\ ] ]
theorem [ theorem:2 ] we adopt the notation @xmath339 for the inverse of @xmath340 with the convention that , whenever we suppress the dependence of @xmath235 on @xmath9 , we mean that it is evaluated at @xmath9 .
moreover , the `` hat '' sign denotes evaluation at @xmath341 as in @xmath342 or in @xmath343 . since @xmath344 , the standardized skewness and kurtosis of @xmath16 are @xmath345 and @xmath346 , where @xmath245 is the standard error of @xmath16 . for @xmath347 and @xmath348 , a cornish - fisher expansion of @xmath349 yields @xmath350 see ( * ? ? ?
* section 4.4 ) .
the following asymptotic expansion of the median of @xmath16 is readily derived : @xmath351 the log - likelihood ratio for @xmath9 is @xmath352 $ ] .
let @xmath353 according to .
it is easy to check that the first three sample derivatives of @xmath109 are @xmath354 $ ] , @xmath355 and @xmath356 .
using the formulae for the derivative of the inverse of a function , one obtains @xmath357 so that @xmath358 and @xmath359 .
let @xmath360 be implicitly defined as a function of @xmath16 and @xmath9 by @xmath361 and assume that @xmath250
. then corresponds to @xmath362 for @xmath363 , cfr . .
next , let @xmath364 be implicitly defined in function of @xmath16 and @xmath9 by @xmath365 .
hence , is implied by @xmath366 cfr . .
we prove by matching the asymptotic expansions of @xmath360 and @xmath364 via an application of lemma 1 and edgeworth expansion of @xmath367 , respectively . as for the former , we define , according to lemma 1 , @xmath368 and @xmath197 by @xmath198 , so that @xmath369 .
one finds @xmath370 and @xmath371 for @xmath202 .
it is easy to show that @xmath372 for any integer @xmath262 , cfr .
, so the hypothesis of lemma 1 are satisfied .
hence an application of lemma [ lemma:1 ] yields @xmath373 for @xmath374/\sigma_\theta = o_p(1)$ ] .
based on , we have @xmath375/\sigma_\theta = ( \theta^{**}-\theta)/\sigma_\theta + \rho_3/6n^{1/2}+o(n^{-3/2})$ ] and @xmath376 . moreover , using a simple taylor expansion , and @xmath372 for any @xmath262 , it can be shown that @xmath377 .
hence , we can reduce to @xmath378 for @xmath363 .
as for the asymptotic expansion of @xmath364 in , let @xmath379 satisfy @xmath380 so that @xmath381 . using cornish - fisher expansion
one finds that , for any @xmath50 , @xmath382 so that @xmath383/3n^{1/2}+o(n^{-3/2})$ ] for @xmath384 .
hence , @xmath385+o(n^{-3/2})\ ] ] for @xmath363 .
next , use the edgeworth expansion for @xmath386 up to the first term , i.e. @xmath387 and a taylor expansion of @xmath388 at @xmath389 for @xmath390 to get @xmath391 substitution into leads to an asymptotic expansion of @xmath392 which corresponds to .
hence , follows and the proof is complete .
theorem [ theorem:2 ] in order to prove , we proceed by deriving two asymptotic expansions for @xmath279 and @xmath277 and by showing that they coincide up to the required order . as for @xmath279 , we resort to equation ( 2.4)(2.6 ) in @xcite . after some algebra and further expansion , @xmath393 so that @xmath394 where we have also used @xmath395 for @xmath396 small . as for @xmath277 ,
a taylor expansion around @xmath9 gives @xmath397 with @xmath398 denoting the remainder .
in the one - parameter exponential family , borrowing the notation from the proof of theorem [ theorem:2 ] , we have @xmath399 since @xmath400 .
moreover , in the proof of theorem [ theorem:2 ] implies that @xmath401 since @xmath402 , @xmath403 and @xmath404 .
inserting and into we obtain the same expansion in provided that the remainder @xmath398 is @xmath405 .
this can be shown by using @xmath406 , @xmath407 , @xmath408 and @xmath409 in the normal deviation range , together with @xmath410 .
hence follows .
[ lemma:1 ] let @xmath411 be a sequence of infinitely differentiable convex functions with minimum at @xmath412 and @xmath413 , and let @xmath197 be defined by @xmath198 . for @xmath414
, assume that , as @xmath41 , @xmath415 for any @xmath202 .
then , @xmath197 admits asymptotic expansion @xmath416 where @xmath417 , @xmath418 and @xmath419 we omit the subscript @xmath45 for ease of notation .
taylor expansion of @xmath420 at @xmath412 gives @xmath421 .
substitute @xmath422 into @xmath423 and equate coefficients of successive order to obtain @xmath424 where the @xmath425 s are positive integers and we set @xmath426 for notational convenience . rearranging terms , the first @xmath427 equations are @xmath428 a similar expression for @xmath429 can be given by means of multinomial coefficients . now substitute back @xmath426 , and solve for @xmath430 to get @xmath431 , @xmath432 and @xmath433 where the order of asymptotics of @xmath434 and @xmath435 are determined by the hypothesis @xmath436 .
an argument by induction leads to .
the authors are grateful to two reviewers for comments that have helped to improve the paper substantially .
special thanks are also due to igor prnster and to mattia ciollaro for comments on an earlier version of this work .
p. de blasi was supported by the european research council ( erc ) through stg `` n - bnp '' 306406 .
barndorff - nielsen , o.e .
( 1983 ) . on a formula for the distribution of the maximum likelihood estimator .
_ biometrika _ * 70 * , 343365 .
barndorff - nielsen , o.e .
inference on full and partial parameters based on the standardized signed log likelihood ratio .
_ biometrika _ * 73 * , 307322 .
barndorff - nielsen , o.e .
approximate interval probabilities . _ j. r. stat .
_ b * 52 * , 485496 .
barndorff - nielsen , o.e . and
cox , d.r .
( 1989 ) . _ asymptotic techniques for use in statistics_. chapman & hall , london .
barndorff - nielsen , o.e . and
cox , d.r .
_ inference and asymptotics_. chapman & hall , london .
birnbaum , a. ( 1961 ) .
confidence curves : an omnibus technique for estimation and testing statistical hypothesis .
_ j. amer .
statist .
_ * 56 * , 246249 .
cox , d. r. ( 1958 ) . some problems with statistical inference .
_ the annals of mathematical statistics _ , * 29 * , 357372 .
efron , b. ( 1982 ) .
transformation theory : how normal is a family of distributions ? _ ann .
statist . _
* 10 * , 323339 .
efron , b. ( 1987 ) .
better bootstrap confidence intervals .
_ j. amer .
statist .
assoc . _ * 82 * , 171185 .
embrechts , p. , klppelberg , c. and mikosch t. ( 1997 ) . _ modelling extremal events for insurance and finance_. springer - verlag berlin heidelberg .
firth , d. ( 1993 ) .
bias reduction of maximum likelihood estimates .
_ biometrika _ * 80 * , 2738 .
fisher , r.a .
inverse probability .
cambridge philos .
* 26 * , 52835 .
lehmann , e.l .
_ testing statistical hypothesis , 2ed_. springer - verlag , new york .
lehmann , e.l . and
casella , g. ( 1998 ) .
_ theory of point estimation , 2ed_. springer - verlag , new york .
neyman , j. ( 1934 ) .
on the two different aspects of the representative method : the method of stratified sampling and the method of purposive selection _ j. r. stat .
_ a * 97 * , 558625 .
schweder , t. ( 2007 ) .
confidence nets for curves . in _ advances in statistical modeling and inference , essay in honour of kjell a. doksum _ , v. nair ed .
world scientific , 593609 .
schweder , t. and hjort , n.l .
confidence and likelihood .
_ scand .
j. statist . _
* 29 * , 309322 .
schweder , t. and hjort , n. l. ( 2016 ) .
_ confidence , likelihood , probability : statistical inference with confidence distributions_. cambridge university press .
skovgaard , i.m .
( 1989 ) . a review of higher order likelihood inference .
inst . _ * 53 * , 331351 .
xie , m. and singh , k. ( 2013 ) .
confidence distribution , the frequentist distribution estimator of a parameter a review .
_ international statistical review _ * 81 * , 3 - 39 . | by the modified directed likelihood , higher order accurate confidence limits for a scalar parameter are obtained from the likelihood .
they are conveniently described in terms of a confidence distribution , that is a sample dependent distribution function on the parameter space . in this paper
we explore a different route to accurate confidence limits via tail
symmetric confidence curves , that is curves that describe equal tailedintervals at any level . instead of modifying the directed likelihood , we consider inversion of the log - likelihood ratio when evaluated at the median of the maximum likelihood estimator .
this is shown to provide equal tailedintervals , and thus an exact confidence distribution , to the third - order of approximation in regular one - dimensional models .
median bias correction also provides an alternative approximation to the modified directed likelihood which holds up to the second order in exponential families .
* keywords : * asymptotic expansion ; confidence curve ; confidence distribution ; exponential family ; modified directed likelihood ; normal transformation family . |
all graphs considered in this paper are finite , undirected , and have neither self - loops nor multiple edges .
a _ graph homomorphism _ from a graph @xmath6 to a graph @xmath7 is a mapping @xmath8 that maps adjacent vertices of @xmath0 to adjacent vertices of @xmath1 , i.e. , @xmath9 whenever @xmath10 .
the notion of a graph homomorphism is well studied in the literature due to its many practical and theoretical applications ; we refer to the textbook of hell and neetil @xcite for a survey .
we write @xmath11 to indicate the existence of a homomorphism from @xmath0 to @xmath1 .
we call @xmath0 the _ guest graph _ and @xmath1 the _ host graph_. we denote the vertices of @xmath1 by @xmath12 and call them _ colors_. the reason for doing this is that graph homomorphisms generalize graph colorings : there exists a homomorphism from a graph @xmath0 to a complete graph on @xmath13 vertices if and only if @xmath0 is @xmath13-colorable
. the problem of testing whether @xmath11 for two given graphs @xmath0 and @xmath1 is called the hom problem .
if only the guest graph is part of the input and the host graph is _ fixed _ ,
i.e. , not part of the input , then this problem is denoted as @xmath1-hom .
the classical result in this area is the hell - neetil dichotomy theorem which states that @xmath1-hom is solvable in polynomial time if @xmath1 is bipartite , and np - complete otherwise @xcite .
we consider so - called _ locally constrained _ homomorphisms .
the _ neighborhood _ of a vertex @xmath14 in a graph @xmath0 is denoted @xmath15 .
if for every @xmath16 the restriction of @xmath17 to the neighborhood of @xmath14 , i.e. , the mapping @xmath18 , is injective , bijective , or surjective , then @xmath17 is said to be _ locally injective _ , _ locally bijective _ , or _ locally surjective _ , respectively . locally bijective homomorphisms are also called _
graph coverings_. they originate from topological graph theory @xcite and have applications in distributed computing @xcite and in constructing highly transitive regular graphs @xcite .
locally injective homomorphisms are also called _ partial graph coverings_. they have applications in models of telecommunication @xcite and in distance constrained labeling @xcite .
moreover , they are used as indicators of the existence of homomorphisms of derivative graphs @xcite .
locally surjective homomorphisms are also called _ color dominations _ @xcite . in addition
they are known as _ role assignments _ due to their applications in social science @xcite .
just like locally bijective homomorphisms they also have applications in distributed computing @xcite .
if there exists a homomorphism from a graph @xmath0 to a graph @xmath1 that is locally bijective , locally injective , or locally surjective , respectively , then we write @xmath19 , @xmath20 , and @xmath21 , respectively .
we denote the decision problems that are to test whether @xmath19 , @xmath20 , or @xmath21 for two given graphs @xmath0 and @xmath1 by lbhom , lihom and lshom , respectively .
all three problems are known to be np - complete when both guest and host graphs are given as input ( see below for details ) , and attempts have been made to classify their computational complexity when only the guest graph belongs to the input and the host graph is fixed .
the corresponding problems are denoted by @xmath1-lbhom , @xmath1-lihom , and @xmath1-lshom , respectively .
the @xmath1-lshom problem is polynomial - time solvable either if @xmath1 has no edge or if @xmath1 is bipartite and has at least one connected component isomorphic to an edge ; in all other cases @xmath1-lshom is np - complete , even when the guest graph belongs to the class of bipartite graphs @xcite .
the complexity classification of @xmath1-lbhom and @xmath1-lihom is still open , although many partial results are known for both problems ; we refer to the papers @xcite and to the survey by fiala and kratochvl @xcite for both np - complete and polynomially solvable cases . instead of fixing the host graph ,
another natural restriction is to only take guest graphs from a special graph class .
heggernes et al .
@xcite proved that lbhom is graph isomorphism - complete when the guest graph is chordal , and polynomial - time solvable when the guest graph is interval .
in contrast , lshom is np - complete when the guest graph is chordal and polynomial - time solvable when the guest graph is proper interval , whereas lihom is np - complete even for guest graphs that are proper interval @xcite .
it is also known that the problems lbhom and lshom are polynomial - time solvable when the guest graph is a tree @xcite . in this paper
we focus on the following line of research .
core _ of a graph @xmath0 is a subgraph @xmath22 of @xmath0 such that @xmath23 and there is no proper subgraph @xmath24 of @xmath22 with @xmath25 .
it is known that the core of a graph is unique up to isomorphism @xcite .
dalmau , kolaitis and vardi @xcite proved that the hom problem is polynomial - time solvable when the guest graph belongs to any fixed class of graphs whose cores have bounded treewidth .
in particular , this result implies an earlier result that hom is polynomial - time solvable when the guest graph has bounded treewidth @xcite .
grohe @xcite strengthened the result of dalmau et al .
@xcite by proving that under a certain complexity assumption , namely @xmath26 $ ] , the hom problem can be solved in polynomial time if and only if this condition holds .
it is a natural question whether the above results of dalmau et al .
@xcite and grohe @xcite remain true when we consider locally constrained homomorphisms instead of general homomorphisms .
we can already conclude from known results that this is not the case for locally surjective homomorphisms . recall that @xmath1-lshom is np - complete even for bipartite guest graphs if @xmath1 contains at least one edge and is either non - bipartite or does not contain a connected component isomorphic to an edge @xcite .
the core of every bipartite graph with at least one edge is an edge , and consequently , has treewidth 1 .
this means that bipartite graphs form a class of graphs whose cores have bounded treewidth . due to this negative answer , we pose the following ( weaker ) question instead : _ are lbhom , lihom and lshom polynomial - time solvable when the guest graph belongs to a class of bounded treewidth ? _ this question
is further motivated by two known results , namely that lbhom and lshom can both be solved in polynomial time if the guest graph is a tree , that is , has treewidth 1 @xcite .
in section [ s - npcom ] , we provide a negative answer to this question by showing that the problems lbhom , lshom and lihom are np - complete already in the restricted case where the guest graph has pathwidth at most @xmath2 , @xmath3 or @xmath4 , respectively .
we also show that the three problems are np - complete even if both the guest graph and the host graph have maximum degree @xmath5 .
the latter result shows that locally constrained homomorphisms problems behave more like unconstrained homomorphisms on graphs of bounded degree than on graphs of bounded treewidth , as it is known that , for example , @xmath27-hom is np - complete on subcubic graphs @xcite . on the positive side , in section [ s - pol ] , we show that all three problems can be solved in polynomial time if we bound the treewidth of the guest graph and at the same time bound the maximum degree of the guest graph or the host graph . because a graph class of bounded maximum degree has bounded treewidth if and only if it has bounded clique - width @xcite , all three problems are also polynomial - time solvable when we bound the clique - width and the maximum degree of the guest graph . in section [ s - pol ]
we also show that lihom can be solved in polynomial time when the guest graph has treewidth 1 , which is best possible given the hardness result for lihom shown in section [ s - npcom ] . in section [ s - con ] we state some relevant open problems .
let @xmath0 be a graph .
the _ degree _ of a vertex @xmath28 in @xmath0 is denoted by @xmath29 , and @xmath30 denotes the maximum degree of @xmath0 .
let @xmath17 be a homomorphism from @xmath0 to a graph @xmath1 .
moreover , let @xmath31 be an induced subgraph of @xmath0 , and let @xmath32 be a homomorphism from @xmath31 to @xmath1 .
we say that @xmath17 _ extends _ ( or , equivalently , is an _ extension _ of ) @xmath32 if @xmath33 for every @xmath34 . a _ tree decomposition _ of @xmath0 is a tree @xmath35 , where the elements of @xmath36 , called the _ nodes _ of @xmath37 , are subsets of @xmath38 such that the following three conditions are satisfied : * for each vertex @xmath39
, there is a node @xmath40 with @xmath41 , * for each edge @xmath10 , there is a node @xmath40 with @xmath42 , * for each vertex @xmath39 , the set of nodes @xmath43 induces a connected subtree of @xmath37 .
the _ width _ of a tree decomposition @xmath37 is the size of a largest node @xmath44 minus one .
the _ treewidth _ of @xmath0 , denoted by @xmath45 , is the minimum width over all possible tree decompositions of @xmath0 .
a _ path decomposition _ of @xmath0 is a tree decomposition @xmath37 of @xmath0 where @xmath37 is a path .
the _ pathwidth _ of @xmath0 is the minimum width over all possible path decompositions of @xmath0 . by definition ,
the pathwidth of @xmath0 is at least as high as its treewidth .
a tree decomposition @xmath37 is _ nice _
@xcite if @xmath37 is a binary tree , rooted in a root @xmath46 such that the nodes of @xmath37 belong to one of the following four types : * a _ leaf node _
@xmath44 is a leaf of @xmath37 , * an _ introduce node _
@xmath44 has one child @xmath47 and @xmath48 for some vertex @xmath49 , * a _ forget node _
@xmath44 has one child @xmath47 and @xmath50 for some vertex @xmath51 , * a _ join node _
@xmath44 has two children @xmath52 satisfying @xmath53 .
for the np - hardness results in theorem [ t - np ] below we use a reduction from the 3-partition problem .
this problem takes as input a multiset @xmath54 of @xmath55 integers , denoted in the sequel by @xmath56 , and a positive integer @xmath57 , such that @xmath58 for all @xmath59 and @xmath60 .
the task is to determine whether @xmath54 can be partitioned into @xmath61 disjoint sets @xmath62 such that @xmath63 for all @xmath64 .
note that the restrictions on the size of each element in @xmath54 implies that each set @xmath65 in the desired partition must contain exactly three elements , which is why such a partition @xmath62 is called a _
@xmath5-partition _ of @xmath54 .
the 3-partition problem is strongly np - complete @xcite , i.e. , it remains np - complete even if the problem is encoded in unary .
[ t - np ] the following three statements hold : * lbhom is np - complete on input pairs @xmath66
where @xmath0 has pathwidth at most @xmath2 and @xmath1 has pathwidth at most @xmath5 ; * lshom is np - complete on input pairs @xmath66 where @xmath0 has pathwidth at most @xmath3 and @xmath1 has pathwidth at most @xmath5 ; * lihom is np - complete on input pairs @xmath66 where @xmath0 has pathwidth at most @xmath4 and @xmath1 has pathwidth at most @xmath4 .
first note that all three problems are in np .
we prove each statement separately starting with statement ( i ) .
and @xmath1 that are constructed from a given instance @xmath67 of @xmath5-partition in the proof of statement ( i ) in theorem [ t - np ] .
see also figure [ fig : detail ] for a more detailed illustration of the `` leftmost '' part of @xmath0 and the `` rightmost '' part of @xmath1 , including more labels . ]
note that lbhom is in np .
given an instance @xmath67 of 3-partition , we construct two graphs @xmath0 and @xmath1 as follows ; see figures [ fig : reduction_bijective ] and [ fig : detail ] for some helpful illustrations .
the construction of @xmath0 starts by taking @xmath55 disjoint cycles @xmath68 of length @xmath57 , one for each element of @xmath54 . for each @xmath59 ,
the vertices of @xmath69 are labeled @xmath70 and we add , for each @xmath71 , two new vertices @xmath72 and @xmath73 as well as two new edges @xmath74 and @xmath75 .
we then add three new vertices @xmath76 , @xmath77 and @xmath78 .
vertex @xmath76 is made adjacent to vertices @xmath79 and @xmath80 for every @xmath59 .
finally , the vertex @xmath77 is made adjacent to every vertex @xmath72 that is not adjacent to @xmath76 , and the vertex @xmath78 is made adjacent to every vertex @xmath73 that is not adjacent to @xmath76 .
this finishes the construction of @xmath0 . to construct @xmath1 ,
we take @xmath61 disjoint cycles @xmath81 of length @xmath57 , where the vertices of each cycle @xmath82 are labeled @xmath83 . for each @xmath64 and @xmath71
, we add two vertices @xmath84 and @xmath85 and make both of them adjacent to @xmath86 .
finally , we add a vertex @xmath87 and make it adjacent to each of the vertices @xmath84 and @xmath85 .
this finishes the construction of @xmath1 . and
@xmath1 in figure [ fig : reduction_bijective ] . ]
we now show that there exists a locally bijective homomorphism from @xmath0 to @xmath1 if and only if @xmath67 is a yes - instance of 3-partition .
let us first assume that there exists a locally bijective homomorphism @xmath17 from @xmath0 to @xmath1 .
since @xmath17 is a degree - preserving mapping , we must have @xmath88 . moreover , since @xmath17 is locally bijective , the restriction of @xmath17 to @xmath89 is a bijection from @xmath89 to @xmath90 . again using the definition of a locally bijective mapping , this time considering the neighborhoods of the vertices in @xmath90
, we deduce that there is a bijection from the set @xmath91 , i.e. , from the set of vertices in @xmath0 at distance @xmath4 from @xmath76 , to the set @xmath92 of vertices that are at distance @xmath4 from @xmath87 in @xmath1 . for every @xmath93
, we define a set @xmath94 such that @xmath95 contains element @xmath96 if and only if @xmath97 .
since @xmath17 is a bijection from @xmath98 to @xmath99 , the sets @xmath62 are disjoint ; moreover each element @xmath96 is contained in exactly one of them .
observe that the subgraph of @xmath0 induced by @xmath98 is a disjoint union of @xmath55 paths of lengths @xmath100 , respectively , while the subgraph of @xmath1 induced by @xmath99 is a disjoint union of @xmath61 cycles of length @xmath57 each .
the fact that @xmath17 is a homomorphism and therefore never maps adjacent vertices of @xmath0 to non - adjacent vertices in @xmath1 implies that @xmath63 for all @xmath64 .
hence @xmath62 is a @xmath5-partition of @xmath54 . for the reverse direction
, suppose there exists a 3-partition @xmath62 of @xmath54 .
we define a mapping @xmath17 as follows .
we first set @xmath101 .
let @xmath102 be any set of the 3-partition .
we map the vertices of the cycles @xmath103 that are at distance @xmath4 from @xmath76 to the vertices of the cycle @xmath82 in the following way : @xmath104 for each @xmath105 , @xmath106 for each @xmath107 , and @xmath108 for each @xmath109 .
the vertices of @xmath110 , @xmath111 and @xmath112 that are at distance more than @xmath4 from @xmath76 in @xmath0 are mapped to vertices of @xmath82 such that the vertices of @xmath110 , @xmath111 and @xmath112 appear in the same order as their images on @xmath82 . in particular ,
we set @xmath104 for each @xmath113 ; the vertices of the cycles @xmath111 and @xmath112 that are at distance more than @xmath4 from @xmath76 are mapped to vertices of @xmath82 analogously .
after the vertices of the cycles @xmath68 have been mapped in the way described above , it remains to map the vertices @xmath72 and @xmath73 for each @xmath59 and @xmath71 .
let @xmath114 be a pair of vertices in @xmath0 that are adjacent to @xmath76 , and let @xmath115 be the second common neighbor of @xmath72 and @xmath73 .
suppose @xmath116 is the image of @xmath115 , i.e. , suppose that @xmath117
. then we map @xmath72 and @xmath73 to @xmath118 and @xmath119 , respectively .
we now consider the neighbors of @xmath77 and @xmath78 in @xmath0 . by construction ,
the neighborhood of @xmath77 consists of the @xmath120 vertices in the set @xmath121 , while @xmath122 .
observe that @xmath87 , the image of @xmath77 and @xmath78 , is adjacent to two sets of @xmath123 vertices : one of the form @xmath118 , the other of the form @xmath119 .
hence , we need to map half the neighbors of @xmath77 to vertices of the form @xmath118 and half the neighbors of @xmath77 to vertices of the form @xmath119 in order to make @xmath17 a locally bijective homomorphism .
the same should be done with the neighbors of @xmath78 . for every vertex @xmath116 in @xmath1
, we do as follows . by construction , exactly three vertices of @xmath0 are mapped to @xmath116 , and exactly two of those vertices , say @xmath115 and @xmath124 , are at distance @xmath4 from @xmath77 in @xmath0 .
we set @xmath125 and @xmath126 .
we also set @xmath127 and @xmath128 .
this completes the definition of the mapping @xmath17 . since the mapping @xmath17 preserves adjacencies
, it clearly is a homomorphism . in order to show that @xmath17 is locally bijective , we first observe that the degree of every vertex in @xmath0 is equal to the degree of its image in @xmath1 ; in particular , @xmath129 . from the above description of @xmath17
we get a bijection between the vertices of @xmath90 and the vertices of @xmath130 for each @xmath131 . for every vertex @xmath72 that is adjacent to @xmath76 and @xmath115 in @xmath0 , its image @xmath118 is adjacent to the images @xmath87 of @xmath76 and @xmath116 of @xmath115 . for every vertex @xmath72 that is adjacent to @xmath77 ( respectively @xmath78 ) and @xmath115 in @xmath0 , its image @xmath118 or @xmath119 is adjacent to @xmath87 of @xmath77 ( respectively @xmath78 ) and @xmath116 of @xmath115 .
hence the restriction of @xmath17 to @xmath132 is bijective for every @xmath59 and @xmath71 , and the same clearly holds for the restriction of @xmath17 to @xmath133 .
the vertices of each cycle @xmath69 are mapped to the vertices of some cycle @xmath134 in such a way that the vertices and their images appear in the same order on the cycles .
this , together with the fact that the image @xmath116 of every vertex @xmath115 is adjacent to the images @xmath118 and @xmath119 of the neighbors @xmath72 and @xmath73 of @xmath115 , shows that the restriction of @xmath17 to @xmath135 is bijective for every @xmath59 and @xmath71 .
we conclude that @xmath17 is a locally bijective homomorphism from @xmath0 to @xmath1 . in order to show
that the pathwidth of @xmath0 is at most @xmath2 , let us first consider the subgraph of @xmath0 depicted on the left - hand side of figure [ fig : detail ] ; we denote this subgraph by @xmath136 , and we say that the cycle @xmath137 _ defines _ the subgraph @xmath136 .
the graph @xmath138 that is obtained from @xmath136 by deleting vertices @xmath139 and edge @xmath140 is a caterpillar , i.e. , a tree in which there is a path containing all vertices of degree more than @xmath141 . since caterpillars are well - known to have pathwidth @xmath141 , graph @xmath138 has a path decomposition @xmath142 of width @xmath141 .
starting with @xmath142 , we can now obtain a path decomposition of the graph @xmath136 by simply adding vertices @xmath76 , @xmath77 , @xmath78 and @xmath143 to each node of @xmath142 ; this path decomposition has width @xmath2 .
every cycle @xmath69 in @xmath0 defines a subgraph @xmath144 of @xmath0 in the same way @xmath137 defines the subgraph @xmath136 .
suppose we have constructed a path decomposition @xmath145 of width @xmath2 of the subgraph @xmath144 for each @xmath59 in the way described above .
since any two subgraphs @xmath144 and @xmath146 with @xmath147 have only the vertices @xmath139 in common , and these three vertices appear in all nodes of each of the path decompositions @xmath145 , we can arrange the @xmath55 path decompositions @xmath148 in such a way that we obtain a path decomposition @xmath149 of @xmath0 of width @xmath2 .
hence @xmath0 has pathwidth at most @xmath2 .
similar but easier arguments can be used to show that @xmath1 has pathwidth at most @xmath5 .
the np - hardness reduction for the locally bijective case can also be used to prove that lihom and lshom are np - hard for input pairs @xmath66 where @xmath0 has pathwidth at most @xmath2 and @xmath1 has pathwidth at most
this follows from the claim that @xmath19 if and only if @xmath21 if and only if @xmath20 for the gadget graphs @xmath0 and @xmath1 displayed in figure [ fig : reduction_bijective ] .
this claim can be seen as follows .
first suppose that @xmath19 .
then , by definition , @xmath21 and @xmath20 .
now suppose that @xmath20 or @xmath21 .
since it can easily be verified that @xmath150 we can use lemma [ l - drm ] ( i ) or ( ii ) , respectively , to deduce that @xmath19 .
however , we can strengthen the hardness results for the locally surjective and injective cases by reducing the pathwidth of the guest graph to be at most @xmath3 and @xmath4 , respectively , and in the latter case we can simultaneously reduce the pathwidth of the host graph to be at most @xmath4 , as claimed in statements ( ii ) and ( iii ) of theorem [ t - np ] . in order to do so , we give the following alternative constructions below . and @xmath151 that are constructed from a given instance @xmath67 of @xmath5-partition in the proof of statement ( ii ) in theorem [ t - np ] . ]
the alternative hardness construction for lshom is similar to but easier than the construction for lbhom ; see figure [ fig : reduction_injective ] .
let @xmath67 be an instance of @xmath5-partition .
we construct a graph @xmath31 by taking @xmath55 disjoint cycles @xmath68 of length @xmath57 , and labeling the vertices of each cycle @xmath69 with labels @xmath152 in the same way as we labeled the vertices of the cycles @xmath69 in the construction for lbhom ( see also figure [ fig : detail ] ) .
we then add two vertices @xmath76 and @xmath77 . for every @xmath59 ,
we make @xmath76 adjacent to each of the vertices @xmath153 , and @xmath77 is made adjacent to each of the vertices @xmath154 .
graph @xmath151 is obtained from the disjoint union of @xmath61 cycles @xmath155 of length @xmath57 by adding one universal vertex @xmath87 . using similar arguments as the ones used in the np - hardness proof of lbhom
, it can be shown that there exists a locally surjective homomorphism @xmath17 from @xmath31 to @xmath151 if and only if @xmath67 is a yes - instance of @xmath5-partition .
such a homomorphism @xmath32 maps @xmath76 and @xmath77 to @xmath87 , and maps the vertices of cycles @xmath68 to the vertices of cycles @xmath155 in exactly the same way as @xmath17 mapped these vertices in the np - hardness proof of lbhom .
it is a routine exercise to show that @xmath31 has pathwidth at most @xmath3 and that @xmath151 has pathwidth at most @xmath5 . and @xmath156 that are constructed from a given instance @xmath67 of @xmath5-partition in the proof of statement ( iii ) in theorem [ t - np ] . ]
the reduction for lihom is even easier ; see figure [ fig : reduction_surjective ] .
given an instance @xmath67 of @xmath5-partition , we create a graph @xmath157 by adding a universal vertex @xmath76 to the disjoint union of @xmath55 paths on @xmath100 vertices , respectively .
graph @xmath156 is obtained from the disjoint union of @xmath61 paths on @xmath57 vertices by adding a universal vertex @xmath87 .
it is easy to verify that there exists a locally injective homomorphism @xmath158 from @xmath157 to @xmath156 , mapping @xmath76 to @xmath87 and all other vertices of @xmath157 to the vertices of degree @xmath4 or @xmath5 in @xmath156 , if and only if @xmath67 is a yes - instance of @xmath5-partition .
the observation that both @xmath157 and @xmath156 have pathwidth @xmath4 completes the proof of theorem [ t - np ] .
we now consider the case where we bound the maximum degree of @xmath0 instead of the treewidth of @xmath0 .
we will combine some known results in order to show that bounding the maximum degree of @xmath0 does not yield tractability for any of our three problems lbhom , lihom and lshom .
we first introduce some additional terminology .
an _ equitable partition _ of a connected graph @xmath0 is a partition of its vertex set in blocks @xmath159 such that any vertex in @xmath160 has the same number @xmath161 of neighbors in @xmath162 .
we call the matrix @xmath163 corresponding to the coarsest equitable partition of @xmath0 ( in which the blocks are ordered in some canonical way ; cf .
@xcite ) the _ degree refinement matrix _ of @xmath0 , denoted as @xmath164 .
we will use the following lemma ; a proof of the first statement in this lemma can be found in the paper of fiala and kratochvl @xcite , whereas the second statement is due to kristiansen and telle @xcite .
[ l - drm ] let @xmath0 and @xmath1 be two graphs .
then the following two statements hold : * if @xmath20 and @xmath165 , then @xmath19 ; * if @xmath21 and @xmath165 , then @xmath19 . kratochvl and kivnek
@xcite showed that @xmath166-lbhom is np - complete , where @xmath166 denotes the complete graph on four vertices . since a graph @xmath0 allows a locally bijective homomorphism to @xmath166
only if @xmath0 is @xmath5-regular , @xmath166-lbhom is np - complete on @xmath5-regular graphs .
the degree refinement matrix of a @xmath5-regular graph is the @xmath167 matrix whose only entry is @xmath5 .
consequently , due to lemma [ l - drm ] , @xmath166-lbhom is equivalent to @xmath166-lihom and to @xmath166-lshom on @xmath5-regular graphs .
this yields the following result .
[ t - degree ] the problems lbhom , lihom and lshom are np - complete on input pairs @xmath168 where @xmath0 has maximum degree @xmath5 .
theorem [ t - degree ] is tight in the following sense .
all three problems lbhom , lihom and lshom are polynomial - time solvable on input pairs @xmath66 where @xmath0 has maximum degree at most @xmath4
. moreover , the first two problems are also polynomial - time solvable on input graphs @xmath66 where only @xmath1 has maximum degree at most @xmath4 .
this does not hold for the lshom , as @xmath169-lshom is np - complete @xcite .
in section [ s - npcom ] , we showed that lbhom , lihom and lshom are np - complete when either the treewidth or the maximum degree of the guest graph is bounded . in this section , we show that all three problems become polynomial - time solvable if we bound both the treewidth and the maximum degree of @xmath0 .
for the problems lbhom and lihom , our polynomial - time result follows from reformulating these problems as constraint satisfaction problems and applying a result of dalmau et al .
@xcite . in order to explain this , we need some additional terminology .
a relational structure @xmath170 is a finite set @xmath54 , called the _ base set _ , together with a collection of relations @xmath171 .
the arities of these relations determine the vocabulary of the structure .
a homomorphism between two relational structures of the same vocabulary is a mapping between the base sets such that all the relations are preserved .
fiala and kratochvl @xcite observed that locally injective and locally bijective homomorphisms between graphs can be expressed as homomorphisms between relational structures as follows .
a locally injective homomorphism @xmath172 can be expressed as a homomorphism between relational structures @xmath173 and @xmath174 , where the new binary relation @xmath175 consists of pairs of distinct vertices that have at least one common neighbor . since @xmath176 maps distinct neighbors of a vertex @xmath28 to distinct neighbors of @xmath177
, we get that @xmath176 is a homomorphism of the associated relational structures . on the other hand ,
if @xmath173 and @xmath174 are constructed from @xmath0 and @xmath1 as described above , and if @xmath176 is a homomorphism between them , then the relations @xmath178 and @xmath179 guarantee that no two vertices with a common neighbor in @xmath0 are mapped to the same target in @xmath1 . in other words , @xmath176 is a locally injective homomorphism between the graphs @xmath0 and @xmath1 . an analogous construction works for locally bijective homomorphisms . here , we need to express @xmath0 using two binary relations @xmath180 and @xmath178 as above , together with @xmath181 unary relations . a unary relation can be viewed as a set : here , the @xmath182-th set will consists of all vertices of degree @xmath183 .
these unary relations guarantee that degrees are preserved , and consequently that the associated graph homomorphisms are locally bijective .
the _ gaifman graph _
@xmath184 of a relational structure @xmath185 is the graph with vertex set @xmath54 , where any two distinct vertices @xmath14 and @xmath28 are joined by an edge if they are bound by some relation .
formally @xmath186 if and only if for some relation @xmath187 of arity @xmath188 and @xmath189 it holds that @xmath190 . as a direct consequence of a result of dalmau et al .
@xcite , the existence of a homomorphism between two relational structures @xmath191 and @xmath192 can be decided in polynomial time if the treewidth of @xmath184 is bounded by a constant .
this leads to theorem [ thm : bounded ] below .
[ thm : bounded ] the problems lbhom and lihom can be solved in polynomial time when @xmath0 has bounded treewidth and @xmath0 or @xmath1 has bounded maximum degree .
first suppose that @xmath0 has bounded treewidth and bounded maximum degree .
observe that for locally injective and locally bijective homomorphisms , the gaifman graph @xmath184 is isomorphic to @xmath193 , which is the graph arising from @xmath0 by adding an edge between any two vertices at distance @xmath4 .
it suffices to observe that @xmath194 , as we can transform any tree decomposition @xmath37 of @xmath0 of width @xmath45 into a desired tree decomposition of @xmath193 by adding to each node @xmath44 of @xmath37 all the neighbors of every vertex from @xmath44 .
since @xmath195 implies that @xmath196 , the theorem also holds if we bound the maximum degree of @xmath1 instead of @xmath0 . to our knowledge ,
locally surjective homomorphisms have not yet been expressed as homomorphisms between relational structures .
hence , in the proof of theorem [ t - bounded ] below , we present a polynomial - time algorithm for lshom when @xmath0 has bounded treewidth and bounded maximum degree .
we first introduce some additional terminology .
let @xmath17 be a locally surjective homomorphism from @xmath0 to @xmath1 .
let @xmath39 and @xmath197 . if @xmath198 , i.e. , if @xmath17 maps vertex @xmath28 to color @xmath199 , then we say that @xmath199 is _ assigned _ to @xmath28 . by definition , for every vertex @xmath39
, the set of colors that are assigned to the neighbors of @xmath28 in @xmath0 is exactly the neighborhood of @xmath200 in @xmath1 .
now suppose we are given a homomorphism @xmath32 from an induced subgraph @xmath31 of @xmath0 to @xmath1 .
for any vertex @xmath34 , we say that @xmath28 _ misses _ a color @xmath197 if @xmath201 , i.e. , if @xmath32 does not assign @xmath199 to any neighbor of @xmath28 in @xmath31 , but any locally surjective homomorphism @xmath17 from @xmath0 to @xmath1 that extends @xmath32 assigns @xmath199 to some neighbor of @xmath28 in @xmath31 .
let @xmath37 be a nice tree decomposition of @xmath0 rooted in @xmath46 .
for every node @xmath40 , we define @xmath202 to be the subgraph of @xmath0 induced by the vertices of @xmath44 together with the vertices of all the nodes that are descendants of @xmath44 .
in particular , we have @xmath203 .
[ def : feasible ] let @xmath40 , and let @xmath204 and @xmath205 be two mappings .
the pair @xmath206 is _ feasible _ for @xmath202 if there exists a homomorphism @xmath17 from @xmath202 to @xmath1 satisfying the following three conditions : * @xmath207 for every @xmath41 ; * @xmath208 for every @xmath41 ; * @xmath209 for every @xmath210 .
in other words , a pair @xmath206 consists of a coloring @xmath211 of the vertices of @xmath44 , together with a collection of sets @xmath212 , one for each @xmath41 , consisting of exactly those colors that @xmath28 misses . informally speaking
, a pair @xmath206 is feasible for @xmath202 if there is a homomorphism @xmath213 such that @xmath17 `` agrees '' with the coloring @xmath211 on the set @xmath44 , and such that none of the vertices in @xmath214 misses any color .
the idea is that if a pair @xmath206 is feasible , then such a homomorphism @xmath17 might have an extension @xmath215 that is a locally surjective homomorphism from @xmath0 to @xmath1 .
after all , for any vertex @xmath41 that misses a color when considering @xmath17 , this color might be assigned by @xmath215 to a neighbor of @xmath28 in the set @xmath216 .
we now prove a result for lshom similar to theorem [ thm : bounded ] .
[ t - bounded ] the problem lshom can be solved in polynomial time when @xmath0 has bounded treewidth and @xmath0 or @xmath1 has bounded maximum degree .
let @xmath66 be an instance of lshom such that the treewidth of the guest graph @xmath0 is bounded . throughout the proof
, we assume that the maximum degree of @xmath1 is bounded , and show that the problem can be solved in polynomial time under these restrictions . since @xmath217 implies that @xmath218 , our polynomial - time result applies also if we bound the maximum degree of @xmath0 instead of @xmath1 .
we may assume without loss of generality that both @xmath0 and @xmath1 are connected , as otherwise we just consider all pairs @xmath219 separately , where @xmath220 is a connected component of @xmath0 and @xmath221 is a connected component of @xmath1 . because @xmath0 has bounded treewidth , we can compute a tree decomposition of @xmath0 of width @xmath45 in linear time using bodlaender s algorithm @xcite .
we transform this tree decomposition into a nice tree decomposition @xmath37 of @xmath0 with width @xmath45 with at most @xmath222 nodes using the linear - time algorithm of kloks @xcite .
let @xmath46 be the root of @xmath37 and let @xmath223 . for each node @xmath40 ,
let @xmath224 be the set of all feasible pairs @xmath206 for @xmath202 .
for every feasible pair @xmath225 and every @xmath41 , it holds that @xmath212 is a subset of @xmath226 . since @xmath227 and @xmath228 for every @xmath41 and every mapping @xmath204
, this implies that @xmath229 for each @xmath40 .
as we assumed that both @xmath13 and @xmath230 are bounded by a constant , the set @xmath224 is of polynomial size with respect to @xmath231 .
the algorithm considers the nodes of @xmath37 in a bottom - up manner , starting with the leaves of @xmath37 and processing a node @xmath40 only after its children have been processed . for every node @xmath44
, the algorithm computes the set @xmath224 in the way described below .
we distinguish between four different cases .
the correctness of each of the cases easily follows from the definition of a locally surjective homomorphism and definition [ def : feasible ] . 1 . _
@xmath44 is a leaf node of @xmath37 .
_ we consider all mappings @xmath204 .
for each mapping @xmath211 , we check whether @xmath211 is a homomorphism from @xmath202 to @xmath1 . if not , then we discard @xmath211 , as it can not belong to a feasible pair due to condition ( i ) in definition [ def : feasible ] . for each mapping
@xmath211 that is not discarded , we compute the unique mapping @xmath232 satisfying @xmath233 for each @xmath41 , and we add the pair @xmath206 to @xmath224 .
it follows from condition ( ii ) that the obtained set @xmath224 indeed contains all feasible pairs for @xmath202 .
as there is no vertex in @xmath214 , every pair @xmath206 trivially satisfies condition ( iii ) .
the computation of @xmath224 can be done in @xmath234 time in this case .
@xmath44 is a forget node .
_ let @xmath47 be the child of @xmath44 in @xmath37 , and let @xmath235 .
observe that @xmath225 if and only if there exists a feasible pair @xmath236 such that @xmath237 and @xmath238 for every @xmath41 , and @xmath239 .
hence we examine each @xmath236 and check whether @xmath239 is satisfied .
if so , we first restrict @xmath240 on @xmath44 to get @xmath206 and then we insert the obtained feasible pair into @xmath224 .
this procedure needs @xmath241 time in total .
@xmath44 is an introduce node .
let @xmath47 be the child of @xmath44 in @xmath37 , and let @xmath242 .
observe that @xmath225 if and only if there exists a feasible pair @xmath236 such that , for every @xmath51 , it holds that @xmath237 , @xmath243 if @xmath10 , and @xmath238 if @xmath244 . hence , for each @xmath236 , we consider all @xmath231 mappings @xmath204 that extend @xmath245 . for each such extension @xmath211 , we test whether @xmath211 is a homomorphism from @xmath202 to @xmath1 by checking the adjacencies of @xmath246 in @xmath1 . if not , then we may safely discard @xmath211 due to condition ( i ) in definition [ def : feasible ] .
otherwise , we compute the unique mapping @xmath205 satisfying @xmath247 and we add the pair @xmath206 to @xmath224 ; due to condition ( ii ) , this pair @xmath206 is the unique feasible pair containing @xmath211 .
computing the set @xmath224 takes at most @xmath248 time in total .
@xmath44 is a join node .
_ let @xmath47 and @xmath249 be the two children of @xmath44 in @xmath37 .
observe that @xmath225 if and only if there exist feasible pairs @xmath250 and @xmath251 such that , for every @xmath41 , @xmath252 and @xmath253 .
hence the algorithm considers every combination of @xmath250 with @xmath251 and if they agree on the first component @xmath211 , the other component @xmath232 is determined uniquely by taking the intersection of @xmath254 and @xmath255 for every @xmath41 .
this procedure computes the set @xmath224 in @xmath256 time in total .
finally , observe that a locally surjective homomorphism from @xmath0 to @xmath1 exists if and only if there exists a feasible pair @xmath206 for @xmath257 such that @xmath258 for all @xmath259 .
since @xmath37 has at most @xmath222 nodes , we obtain a total running time of @xmath260 . as we assumed that both @xmath223 and @xmath230 are bounded by a constant
, our algorithm runs in polynomial time .
note that theorem [ thm : bounded ] can be derived by solving lihom using a dynamic programming approach that strongly resembles the one for lshom described in the proof of theorem [ t - bounded ] , together with the fact that @xmath66 is a yes - instance of lbhom if and only if it is a yes - instance for both lihom and lshom . in a dynamic programming algorithm for solving lihom , instead of keeping track of sets @xmath212 of colors that a vertex @xmath41 is missing , we keep track of sets @xmath261 of colors that have already been assigned to the neighbors of a vertex @xmath41 . this is because in a locally injective homomorphism from @xmath0 to @xmath1 , no color may be assigned to more than one neighbor of any vertex . in this way
we can adjust definition [ def : feasible ] in such a way that it works for locally injective instead of locally surjective homomorphisms .
we omit further details , but we expect that a dynamic programming algorithm of this kind will have smaller hidden constants in the running time estimate than the more general method of dalmau et al .
@xcite .
we conclude this section with one more polynomial - time result .
it is known that the problems lbhom and lshom are polynomial - time solvable when @xmath0 is a tree @xcite , and consequently when @xmath0 has treewidth @xmath141 .
we claim that the same holds for the lihom problem .
[ t - polytree ] the lihom problem can be solved in polynomial time when @xmath0 has treewidth @xmath141 .
let us first state some terminology and useful known results .
the universal cover @xmath262 of a connected graph @xmath0 is the unique tree ( which may have an infinite number of vertices ) such that there is a locally bijective homomorphism from @xmath262 to @xmath0 .
one way to define this mapping is as follows .
consider all finite walks in @xmath0 that start from an arbitrary fixed vertex in @xmath0 and that do not traverse the same edge in two consecutive steps .
each such walk will correspond to a vertex of @xmath262 .
we let two vertices of @xmath262 be adjacent if and only if one can be obtained from the other by deleting the last vertex of the walk .
then the mapping @xmath263 that maps every walk to its last vertex is a locally bijective homomorphism from @xmath262 to @xmath0 @xcite . it is also known that @xmath264 if and only if @xmath0 is a tree @xcite . moreover , for any two graphs @xmath0 and @xmath1 , @xmath20 implies that @xmath265 @xcite .
now let @xmath66 be an instance of lihom where @xmath0 has treewidth @xmath141 .
we assume , without loss of generality , that both @xmath0 and @xmath1 are connected .
in particular , @xmath0 is a tree .
we claim that @xmath20 if and only if @xmath265 . the forward implication follows from above . to show the backward implication , suppose that @xmath265 .
then @xmath266 , because @xmath264 .
let @xmath176 be a locally injective homomorphism from @xmath0 to @xmath267 . then , because @xmath266 and @xmath268 , we have @xmath20 .
to explain this , consider the mapping @xmath269 defined by @xmath270 if and only if @xmath271 .
notice that @xmath272 is a locally injective homomorphism from @xmath0 to @xmath1 .
the desired result follows from this claim combined with the fact that we can check in polynomial time whether @xmath265 holds for two graphs @xmath0 and @xmath1 @xcite .
theorem [ t - polytree ] states that lihom can be solved in polynomial time when the guest graph has treewidth 1 , while theorem [ t - np ] implies that the problem is np - complete when the guest graph has treewidth 2 .
this shows that the bound on the pathwidth in the third statement of theorem [ t - np ] is best possible .
we leave it as an open problem to determine whether the bounds on the pathwidth in the other two statements of theorem [ t - np ] can be reduced further .
we conclude this paper with some remarks on the parameterized complexity of the problems lihom , lshom and lbhom .
the hardness results in this paper show that all three problems are para - np - complete when parameterized by either the treewidth of @xmath0 or the maximum degree of @xmath0 .
theorems [ thm : bounded ] and [ t - bounded ] show that the problems are in xp when parameterized jointly by the treewidth of @xmath0 and the maximum degree of @xmath0 .
a natural question is whether the problems are fpt when parameterized by the treewidth of @xmath0 and the maximum degree of @xmath0 , i.e. , whether they can be solved in time @xmath273 for some function @xmath176 that does not depend on the sizes of @xmath0 and @xmath1 .
we would like to thank isolde adler for posing the research questions that we addressed in our paper and for helpful discussions .
the fourth author also thanks jan arne telle for fruitful discussions . | a homomorphism from a graph @xmath0 to a graph @xmath1 is locally bijective , surjective , or injective if its restriction to the neighborhood of every vertex of @xmath0 is bijective , surjective , or injective , respectively .
we prove that the problems of testing whether a given graph @xmath0 allows a homomorphism to a given graph @xmath1 that is locally bijective , surjective , or injective , respectively , are np - complete , even when @xmath0 has pathwidth at most @xmath2 , @xmath3 , or @xmath4 , respectively , or when both @xmath0 and @xmath1 have maximum degree @xmath5 .
we complement these hardness results by showing that the three problems are polynomial - time solvable if @xmath0 has bounded treewidth and in addition @xmath0 or @xmath1 has bounded maximum degree . computational complexity ; locally constrained graph homomorphisms ; bounded treewidth ; bounded degree |
let @xmath1 be a bounded and measurable function in @xmath2 such that the support of @xmath3 is a bounded domain @xmath4 .
we are concerned with the acoustic scattering problem @xmath5 where @xmath6 and @xmath7 is an incident field satisfying @xmath8 in @xmath2 . for simplicity , we take incident plane waves @xmath9 where @xmath10 and @xmath11 is the unit sphere .
the scattered field @xmath12 satisfies the sommerfeld radiation condition : @xmath13 the scattering problem ( [ acimpoenetrable-2]-[radiationc-2 ] ) is well posed , see @xcite . applying green s formula to @xmath12 , we can show that the scattered field @xmath14 has the following asymptotic expansion : @xmath15 where the function @xmath16 for @xmath17 is the corresponding far - field pattern .
our interest in this work is related to the classical inverse scattering problem which consists of reconstructing @xmath18 from the far - field data @xmath16 for some @xmath17 .
this type of problem is well studied and there are several algorithms to solve it in the case when @xmath19 and @xmath20 are taken in the whole @xmath21 , see @xcite .
it is also known that this problem is very unstable .
precisely , the modulus of continuity is in general of logarithmic type , see @xcite .
recently , based on a new type of experiments , a different approach was proposed , see for instance @xcite and the references therein .
it is divided into two steps . in the first one , we deform the acoustic medium by small inclusions located in the region containing the support of @xmath3 , and
measure the far - fields generated by these deformations . from these far - fields
, we can extract the total fields due to the medium @xmath1 in the interior of the support of @xmath3 . in the second step ,
we reconstruct @xmath22 from these interior values of the total fields .
recently , there was an increase of interest in reconstructing media from internal measurements .
this is also related to the hybrid methods introduced in the medical imaging community , see @xcite for different setups and models .
in contrast to the instability of the classical inverse scattering problem , the reconstruction from internal measurements is stable , see @xcite . however , the disadvantage with internal fields , i.e. internal values of the total fields , is the existence of their zeros which need to be properly dealt with to stabilize any algorithm . to overcome this short coming ,
one can think of using measurements related to multiple frequencies @xmath23 , see @xcite .
our objective in this paper is to propose an alternative to overcome the last disadvantage . for this
, we propose to deform the medium using multiple ( precisely double ) and close inclusions instead of just single inclusions .
then , measuring the generated far - fields by these multiple inclusions , we can extract , not only the internal total fields , but also the internal values of the green s function related to the non deformed medium @xmath1 .
this is possible because by deforming the medium with multiple and close inclusions we generate scattered fields due to the multiple scattering between these inclusions .
precisely , the far - fields we measure encode at least the second order term in the foldy - lax approximation and not only the first order ( or the born approximation ) as it is done when deforming with single or multiple but well separated inclusions .
finally , we extract the values @xmath22 , @xmath24 in @xmath4 , from the singularities of this green s function .
hence , we avoid the problems coming from the zeros of the internal total fields .
the accuracy of the reconstruction is related to the minimum distance @xmath25 between the pairs of inclusions , in addition to the radius of the perturbations @xmath26 . in practice ,
the perturbations are created using focusing waves , see @xcite for instance , hence we are limited by the resolution of the used waves ( acoustic waves for instance ) .
in other words , the small perturbations can not be very close . in our analysis ,
the ratio @xmath27 is of the form @xmath28 , with a parameter @xmath29 $ ] , and it can be chosen small by taking @xmath30 near @xmath31 . this allows to have the minimum distance between the inclusions quite large compared to their sizes .
since the accuracy of the reconstruction , see theorem [ main - results ] , is of the order @xmath32 , with @xmath30 in @xmath33 $ ] with some @xmath34 , then one should find a reasonable balance between the limited resolution of the focusing waves , to be used to perturb the medium , and the desired accuracy of the reconstruction .
more details related to this issue are provided in remark [ remark ] .
we complete this section by describing the kind of deformations and the corresponding far - field measurements we use and then we state the derived formulas to extract the values of the index of refraction @xmath1 . in section @xmath35 , we justify these formulas with the emphasize on the explicit dependence of the error terms on the parameters modeling the small perturbations ( i.e. the maximum radii , the minimum distance between them and the scaled surface impedances ) . in section @xmath36 , we discuss the stability issue by providing the corresponding formulas when the measured data are contaminated with additive noises and when the small perturbations are shifted from their exact locations ( as moving from step @xmath35 to step @xmath36 in collecting the data , see section @xmath37 ) .
we provide the regimes under which the derived formulas are still valid . we give the details of this approach by using inclusions of the form of obstacles of impedance type .
let @xmath38 be the ball with the center at the origin and radius @xmath39 .
we set @xmath40 to be the small bodies characterized by the parameter @xmath41 and the locations @xmath42 , @xmath43 .
we denote by @xmath44 the acoustic field scattered by the @xmath45 small bodies @xmath46 , due to the incident field @xmath47 ( mainly the plane incident waves @xmath48 with the incident direction @xmath10 , where @xmath11 being the unit sphere ) , with impedance boundary conditions .
hence the total field @xmath49 satisfies the following exterior impedance problem for the acoustic waves @xmath50 @xmath51 @xmath52 the scattering problem ( [ acimpoenetrable-1])-([radiationc-1 ] ) is well posed , see @xcite , and we can also allow @xmath53 to be negative , see @xcite . [ def1 ] we define 1 .
@xmath54 where @xmath55 .
2 . @xmath56 as the upper bound of the used wave numbers , i.e. @xmath57 $ ] . + the distribution of the scatterers is modeled as follows : 3 .
the number @xmath58 with a given positive constant @xmath59 .
the minimum distance @xmath60 , i.e. @xmath61 , with given positive constants @xmath62 and @xmath63 .
the surface impedance @xmath64 , where @xmath65 and might be a complex number . here
the real numbers @xmath66 , @xmath30 and @xmath67 are assumed to be non negative .
we call @xmath68 and @xmath69 the set of the a priori bounds . in ( @xcite ,
corollary 1.3 ) , we have shown that there exist positive constants @xmath70 , @xmath71 , @xmath72 , we need no a priori conditions on @xmath71 , @xmath72 . ] depending only on the set of the a priori bounds and on @xmath73 such that if @xmath74 then the far - field pattern @xmath75 has the following asymptotic expansion @xmath76 uniformly in @xmath19 and @xmath20 in @xmath11 . here
the constant appearing in the estimate @xmath78 depends only on the set of the a priori bounds , @xmath71 , @xmath72 and on @xmath79 .
the coefficients @xmath80 , @xmath81 are the solutions of the following linear algebraic system @xmath82 for @xmath83 where @xmath84 and @xmath85 is the green function corresponding to the scattering problem ( [ acimpoenetrable-2]-[radiationc-2 ] ) .
the quantity @xmath86 is a constant if @xmath87 is a ball and this constant is equal to the radius of @xmath87 , i.e. @xmath88 . the algebraic system is invertible under the condition : @xmath89 we use surface impedance functions of the form @xmath90 , i.e. with @xmath91 in the case @xmath92 , so that the constants @xmath93 are well defined . in the subsequent sections ,
we choose @xmath94 and @xmath95 , @xmath96 . in this case
@xmath97 hence ( [ x oustdie1 d_m farmain-2 ] ) becomes @xmath98 uniformly in @xmath19 and @xmath20 in @xmath11 .
we proceed in three steps in collecting the measured data : 1 .
first step .
we measure the far - fields before making any deformation . in this case
the collected data are @xmath99 , @xmath100 .
2 . second step .
we deform the medium by , and then measure the corresponding data @xmath101 , for the directions @xmath102 , @xmath103 , keeping in mind that we redo the experiment by moving the inclusion centered at @xmath104 for @xmath105 and @xmath106 , see fig [ fig:1 ] .
3 . third step .
we deform the medium using two inclusions close to each other i.e. the two points @xmath107 and @xmath108 in @xmath4 , are such that @xmath109 , as @xmath110 , with @xmath111 , see fig [ fig:2 ] . in this case , the collected data are @xmath112 , @xmath113 . as in step @xmath35 , we create one couple of two close inclusions at once .
then , we move the couple of inclusions , for @xmath114 , inside @xmath4
. the reconstruction formulas of the index of refraction from the described measured data are summarized in the following theorem .
[ main - results ] assume @xmath1 to be a measurable and bounded function in @xmath2 such that the support of @xmath3 is a bounded domain @xmath4 .
let the non negative parameters @xmath115 and @xmath30 describing the set of the small inclusions , embedded in @xmath4 , be such that @xmath116 let @xmath105 be fixed .
we have the following formulas : 1 .
the vector of total fields @xmath117 ( or @xmath118 ) can be recovered from @xmath119 as follows : @xmath120=-v^t(z^l_m , \theta_{i})v^t(z^l_m , \theta_{j})+o(a),\ ] ] @xmath121 . ,
see ( [ actual - capacitances ] ) .
] as the two points @xmath107 and @xmath108 are close then the sign before the vectors above is the same for @xmath122 and @xmath123 .
2 . knowing the vectors of total fields @xmath117 , for @xmath106 , ( or @xmath118 , for @xmath106 ) the values of the green function @xmath124 can be recovered as follows .
we have @xmath125 where the matrix @xmath126 is computable as @xmath127\textbf{v}^\top(\textbf{v}\textbf{v}^\top)^{-1 } \textcolor{black}{+ } \textbf{c}^{-1}=\tilde{\textbf{g}}+\;~ o\left ( a^{1-h-2t}+a^{h}\right)\ ] ] with the matrix @xmath128 constructed from ( [ total - fields ] ) via @xmath129 , @xmath106 and @xmath130 and @xmath131 .
is diagonal and the diagonal values are defined by the vector @xmath132 . since @xmath133 as the shapes of the inclusions are the same , i.e. balls of radius @xmath26 , we have @xmath134 where @xmath135 is the identity matrix in @xmath136 . ] the matrices @xmath137 and @xmath138 are defined as @xmath139 and @xmath140 , for @xmath141 .
3 . if , in addition , @xmath1 is of class @xmath142,$ ] then , i.e. @xmath1 is only continuous , and obtain ( [ formulas - for - n ] ) with an appropriate change in the error term . ] from the values of the green function @xmath124 , we reconstruct the values of the index of refraction @xmath143 as follows : @xmath144 + o(a^t),\;~ t>0.\ ] ] [ remark ] : 1 .
we observe that due to the error @xmath145 , for @xmath146 , we do not distinguish between @xmath147 and @xmath148 since @xmath149 , as @xmath150 and hence @xmath151 ( in the best case @xmath152 ) .
hence these formulas provide the values of @xmath153 , for @xmath146 , where @xmath154 for either @xmath122 or @xmath123 .
2 . from the formulas ( [ total - fields ] ) ,
( [ green - function-2 ] ) and ( [ formulas - for - n ] ) , we see that the error in reconstructing @xmath1 is of the order @xmath155 and hence we need @xmath156 and @xmath157 .
the ratio between the radius @xmath26 and the minimum distance @xmath158 is of the order @xmath159 as @xmath160 .
this rate becomes significantly smaller when @xmath30 approaches @xmath31 .
this means that the small perturbations do not need to be very close to each other to apply the reconstruction formulas in theorem [ main - results ] .
this makes sense and can be useful in practice since , due to the resolution limit , it is delicate to use focusing waves to create close small deformations .
the price to pay is that we will have less accuracy in the reconstruction , see the formula ( [ formulas - for - n ] ) for instance .
hence , one needs to find a compromise between the desired accuracy of the reconstruction and the permissible closeness of the small perturbations created by focusing waves .
for instance taking @xmath161 , the error in the reconstruction is of the order @xmath162 and the ratio @xmath27 is of the order @xmath163 , as @xmath160 .
the importance of the parameter @xmath30 , describing the closeness of the small inclusions , appears in the last step ( i.e. the point 3 . in theorem [ main - results ] ) to compute the values of @xmath1 using the singularity analysis of .
indeed , in ( [ green - function-2 ] ) , we can even take @xmath164 with @xmath165 . if we need to choose @xmath30 very small ( for practical or experimental reasons ) , then we can also use other methods to compute @xmath22 from @xmath166 .
for instance , from the lippmann - schwinger equation @xmath167 , we can compute @xmath168 by inverting an integral equation of the first kind having as a kernel @xmath169 .
knowing @xmath166 and @xmath170 for a sample of points @xmath171 , we can reconstruct @xmath1 from the formula @xmath172 .
the denominator @xmath173 is not vanishing if we choose appropriately different values of the source points @xmath0 in @xmath4 .
4 . in this work ,
we use localized perturbations modeled by small inclusions with impedance boundary conditions .
however , the natural localized perturbations should occur on the coefficients of the model @xcite , i.e. the index of refraction @xmath1 in our model . in this respect , using surface impedance type inclusions might be more attractive in the academic level than in practice unless we accept and use the following argument . in case where
there is a large contrast between the medium and the background , and according to the leontovich approximation , the contrast can be replaced by the surface impedance boundary condition .
this means that we should use localized perturbations creating quite high contrasts .
we create one single inclusion located at a selected point @xmath104 in @xmath4 .
this means that we take @xmath175 in the asymptotic expansion ( [ x oustdie1 d_m farmain-2 ] ) , i.e. @xmath176 and @xmath177 .
hence @xmath178 using the backscattered data @xmath179 and @xmath180 , and since @xmath181 is known , from ( [ m-1 ] ) we derive the formula @xmath182=-(v^t(z^l_m , \theta_j))^2+\ ; o\left ( a\right).\ ] ] hence we can construct either @xmath183 or @xmath184 since the complex argument will be computed to an additive @xmath185 period . from ( [ m-1 ] ) , we see that we can compute @xmath186 , @xmath187 , with same additive period
. then , we can construct either the vector @xmath117 or the vector @xmath118 .
let us now show why if @xmath107 is close to @xmath108 , then this sign should be the same for @xmath122 and @xmath123 .
we know that the total fields @xmath188 are solutions to the lippmann - schwinger equations @xmath189 based on this equation and following the arguments in lemma 2.2 of @xcite , we show that @xmath190 is in @xmath191 with the bound @xmath192 where @xmath193 depends only on the upper bound of @xmath1 and @xmath194 , here @xmath195 . as @xmath20 varies in @xmath11 , the constant @xmath196 is uniformly bounded .
we set @xmath197 .
hence we deduce that @xmath198 hence if the points @xmath107 and @xmath108 are chosen such that @xmath199 is small enough , then @xmath200 and @xmath201 will have the same sign and @xmath202 fixed , the value @xmath183 can vanish for some @xmath203 but according to lemma [ matrix - v - inver ] it can not be true for ever @xmath204 where @xmath205 is large . ] . in this case , we create two inclusions located close to each other , i.e. the two points @xmath107 and @xmath108 , @xmath114 , in @xmath4 , are such that @xmath109 , as @xmath110 , with @xmath111 .
let @xmath207 be fixed . from the asymptotic expansion ( [ x oustdie1 d_m farmain-2 ] ) , we have @xmath208 where @xmath209 hence , we rewrite ( [ m - finite ] ) as @xmath210 where we use the vector @xmath211^\top$ ] and the @xmath212 matrix @xmath213 whose components are @xmath214 for @xmath215 and @xmath216 for @xmath217 .
we use a number @xmath218 of directions of incidences ( and hence of propagation directions ) larger than the number of sampling points @xmath219 , i.e. @xmath35 directions , where we want to evaluate the index of refraction . with this number at hand , the matrix @xmath220\ ] ] has a full rank and hence @xmath221 is invertible .
precisely , we have the following lemma .
[ matrix - v - inver ] there exists @xmath222 such that if the number @xmath218 of incident directions @xmath223 is larger than @xmath205 , i.e. @xmath224 , then the vectors @xmath225^\top$ ] , @xmath226 are linearly independent .
we observe , by the mixed reciprocity relations , that @xmath227 , where @xmath228 is the far - field in the direction @xmath229 of the point source @xmath230 supported at the source point @xmath104 .
we recall that the point source @xmath230 is the green function of the model ( [ acimpoenetrable-2 ] ) . with this relation at hand , the proof of
the lemma follows the same arguments in @xcite where it is done for the vectors @xmath231 and @xmath232 , with @xmath218 large enough .
we omit the details .
observe that @xmath233 if the index of refraction @xmath1 is equal to unity in @xmath2 . in practice
, one may take @xmath234 .
this can be easily justified for the case when the contrast of the index of refraction , @xmath3 , is small .
to see it we set , keeping the same notation as above , @xmath235 $ ] where @xmath236^\top$ ] , @xmath237 . correspondingly , we set @xmath238^\top$ ] , @xmath239 and @xmath240 $ ] .
a simple computation shows that the determinant of @xmath241 is @xmath242\neq 0 $ ] if @xmath243 .
now , we observe , by the lippmann - schwinger equation for instance , that @xmath244 . hence if the contrast @xmath245 is small enough then the determinant of @xmath246 is not vanishing .
hence , by measuring the data @xmath247 and @xmath248 using the formulas@xmath249 and @xmath250 we obtain the formula @xmath251 observe that @xmath252 where @xmath253 and @xmath254 and @xmath255 .
hence we can write @xmath256 where @xmath257 is the @xmath258-norm for matrices .
we recall that we have chosen the points @xmath107 and @xmath108 , in @xmath4 , such that @xmath109 with @xmath259 , as @xmath110 .
with such a choice , we see that @xmath260 as @xmath261 .
finally , from the matrix @xmath262 we derive the values of the green function at the sample points @xmath263 , in @xmath4 : @xmath264 indeed , combining ( [ matrix - b - q ] ) and ( [ first - order - appr-1 ] ) , we deduce that @xmath265 where @xmath266 and @xmath267 .
remember that from step .
@xmath39 , we compute @xmath268 ( i.e. we do know the correct sign ) , but as in ( [ approximation ] ) we multiply the two sides by @xmath246 and @xmath269 , then the missed sign of @xmath246 is irrelevant .
the approximation ( [ approximation ] ) will make sense , i.e. the error term is dominated by @xmath270 , if @xmath271 and @xmath272 , i.e. @xmath273 .
in addition , @xmath274 if @xmath275 , i.e. @xmath276 . hence the approximation ( [ approximation ] ) makes sense if @xmath277 . this last condition is satisfied since we assumed that @xmath278 and @xmath279 .
from step 1 , we showed already how to extract the matrices @xmath268 , hence we can compute @xmath280 .
since the matrix @xmath281 is diagonal , with non zero entries , and the matrix @xmath282 is invertible , as @xmath128 is full rank matrix , then we can recover the matrix @xmath126 as @xmath283\textbf{v}^\top(\textbf{v}\textbf{v}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}=\tilde{\textbf{g}}+\;~ o(a^{2h-2})\;o\left ( a^{3 - 3h-2t}+a^{2-h}\right)\ ] ] or @xmath284\textbf{v}^\top(\textbf{v}\textbf{v}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}=\tilde{\textbf{g}}+\;~ o\left ( a^{1-h-2t}+a^{h}\right).\ ] ] we see that @xmath285 if @xmath286,i.e .
@xmath287 and @xmath288 if @xmath289 , i.e. @xmath290 . hence ( [ reconvering - g - kappa-1 ] ) makes sense if @xmath291 .
as mentioned above , this last condition is satisfied since we assumed that @xmath278 and @xmath279 .
finally , we set @xmath292 as the reconstructed values of the green function @xmath293 at the sampling points .
we start with the following lemma we have the asymptotic expansion : @xmath294 we know that @xmath295 in @xmath2 and @xmath296 satisfies @xmath297 in @xmath2 , where both @xmath293 and @xmath298 satisfy the sommerfeld radiation condition .
then @xmath299 satisfies the same radiation condition and @xmath300 multiplying both sides of ( [ difference ] ) by @xmath298 and integrating over a bounded and smooth domain @xmath38 we obtain : @xmath301 and then @xmath302 here , we choose @xmath38 such that @xmath303 .
since @xmath304 , as the singularity of @xmath305 is of the order @xmath306 , and @xmath307 , since @xmath1 is of class @xmath308 , we deduce that @xmath309 since @xmath310 , see ( @xcite , lemma 4.1 ) , for instance , for the such integrals o f of singular functions .
the two last integrals appearing in ( [ representation-2 ] ) are of the order @xmath311 for @xmath312 .
now , let @xmath313 be the ball of center @xmath314 and radius @xmath315 .
we use ( [ representation-1 ] ) applied to @xmath313 instead of @xmath38 : @xmath316 for @xmath317 .
we see that @xmath318 which we can rewrite as @xmath319 for @xmath317 .
hence @xmath320 since @xmath321 is bounded for @xmath322 , then @xmath323 for @xmath324 .
hence , we can write ( [ representation-3 ] ) as @xmath325 taking the trace in ( [ representation-4 ] ) to @xmath326 and using the trace of the double layer potential @xmath327 , where @xmath328 denotes the double layer operator , we derive the equation @xmath329 then @xmath330 we have the following estimate from @xcite : @xmath331 where @xmath332 . hence @xmath333 going back to ( [ representation-4 ] ) , we estimate @xmath334 and then @xmath335 from ( [ abs - w ] ) and the fact that @xmath336 , we derive the estimate @xmath337 for @xmath338 and in particular for @xmath24 such that @xmath339 .
hence , from the definition of @xmath340 , we deduce that @xmath341 from ( [ asymp - g ] ) , we deduce that , @xmath342 and then @xmath343.\ ] ] the formula ( [ final - formula-1 ] ) can be used for @xmath344 and @xmath345 , where @xmath346 and @xmath108 , are the sampling points added to the small inclusions , @xmath347 + o(a^t),\;~ t>0.\ ] ]
the object of this section is to analyze how the formulas are accurate when the measured far - fields are contaminated with noise and when the locations of the small perturbations @xmath348 , are slightly shifted when passing from step 2 to step 3 in collecting the measured far - fields .
precisely , we will derive the regimes on the amplitude of the noise and the length of the shifts under which our formulas are still applicable .
we recall the main formulas , and ; @xmath349\textbf{v}^\top(\textbf{v}\textbf{v}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}=\tilde{\textbf{g}}+\;~ o\left ( a^{1-h-2t}+a^{h}\right).\ ] ] @xmath350 and @xmath351 + o(a^t),\;~ t>0.\ ] ] combining them , we obtain : @xmath352\textbf{v}^\top ( \textbf{v}\textbf{v}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}\right]_{_{1,2 } } -\frac{1}{\;4\pi\vert z^2_m - z^1_m\vert}\quad\\ & & + o\left ( a^{1-h-2t}+a^{h}\right)+o\left(a^{t}\right).\nonumber \end{aligned}\ ] ] we recall also that where @xmath211^\top$ ] and @xmath353 can be computed , see and , as @xmath354&=&-v^t(z^l_m , \theta_i)v^t(z^l_m , \theta_j)+\ ; o\left(a\right).\end{aligned}\ ] ] we base our remaining analysis on the two formulas ( [ true - formula ] ) and ( [ one - inclusion - stab ] ) .
we set @xmath355 to be the possibly shifted positions of the small perturbations while moving from the second step to third step of collecting the far - fields , see section [ sec : extractform ] .
we set also @xmath356 and @xmath357 where @xmath358 s are the ball of center @xmath359 s and radius @xmath26 respectively .
let now @xmath360 and @xmath361 be the additive noises corresponding to @xmath362 and @xmath363 respectively .
the actual measured data is @xmath364 and @xmath365 satisfying respectively @xmath366 uniformly in terms of @xmath367 , @xmath368 .
observe that since @xmath369=-v^t(z^l_m , \theta_i)v^t(z^l_m , \theta_j)+\ ; o\left(a\right)\nonumber $ ] then @xmath370=-v^t(z^l_m , \theta_i ) v^t(z^l_m , \theta_j)+\ ; o\left(a\right)+c^{-1}_{m , l } ( \delta_u+\delta_v).\end{aligned}\ ] ] we set where @xmath371^\top$ ] .
remember now that in the second step , the data are collected at the shifted points .
hence instead of ( [ true - formula ] ) , we have @xmath372\tilde{\textbf{v}}^\top ( \tilde{\textbf{v}}\tilde{\textbf{v}}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}\right]_{_{1,2 } } -\frac{1}{\;4\pi\vert \tilde{z}^2_m-\tilde{z}^1_m\vert } \quad\\ & & + o\left ( a^{1-h-2\tilde{t}}+a^{h}\right)+o\left ( a^{\tilde{t}}\right)\nonumber\end{aligned}\ ] ] where we set @xmath373 to be the matrix defined as @xmath374 and correspondingly we set the matrix @xmath375 defined as @xmath376 . here
@xmath377 satisfies the conditions @xmath378 from this formula , we get the corresponding one with noisy data @xmath379 and @xmath380 @xmath381\tilde{\textbf{v}}^\top ( \tilde{\textbf{v}}\tilde{\textbf{v}}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}\right]_{_{1,2 } } -\frac{1}{\;4\pi\vert \tilde{z}^2_m-\tilde{z}^1_m\vert } \qquad\\ & & + ( \delta_w+\delta_v)\vert \textbf{c}^{-1}(\tilde{\textbf{v}}\tilde{\textbf{v}}^\top)^{-1 } \tilde{\textbf{v}}\vert \vert \tilde{\textbf{v } } ( \tilde{\textbf{v}}\tilde{\textbf{v}}^\top)^{-1 } \textbf{c}^{-1}\vert+\;~ o\left ( a^{1-h-2\tilde{t}}+a^{h}\right)+o\left ( a^{\tilde{t}}\right).\nonumber\end{aligned}\ ] ] now , we need to replace the values of @xmath382 by the values of @xmath128 , more precisely the computed values of @xmath128 from step 1 with noisy data .
let us set @xmath383 the computed one from step 1 .
we deduce then from ( [ upper - bound ] ) and ( [ v - with - noisy - data ] ) that @xmath384 with this estimate at hand , the final formula is : @xmath385\textbf{v}_{\delta}^\top ( \textbf{v}_{\delta}\textbf{v}_{\delta}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}\right]_{_{1,2 } } -\frac{1}{\;4\pi\vert \tilde{z}^2_m-\tilde{z}^1_m\vert}\ ] ] @xmath386 @xmath387 as @xmath388 and from ( [ approximation ] ) , applied to the points @xmath355 s instead of the points @xmath104 s , we have @xmath389 , then @xmath390 what is left is the estimate of the term @xmath391 ( respectively @xmath392 ) or the term @xmath393 ( respectively @xmath394 ) since the matrix @xmath395 . arguing as in lemma [ matrix - v - inver ]
, the matrix @xmath396 is invertible .
however , what we need here is its lower bound .
wet set @xmath397 its lower bound
. then @xmath398 . with these estimates ,
the error term in ( [ true - formula - shifted-2 ] ) becomes @xmath399 or @xmath400 @xmath401 if we reasonably assume that we have uniform noise for our different measurements , i.e. @xmath402,@xmath403 and @xmath404 , then this reduces to @xmath405 we summarize these findings in the following theorem assume @xmath1 to be a measurable and bounded function in @xmath2 such that the support of @xmath3 is a bounded domain @xmath4 .
let the non negative parameters @xmath115 and @xmath30 describing the set of the small inclusions , embedded in @xmath4 , be such that @xmath406 we assume that the measured far - fields are contaminated with an additive noise of amplitude @xmath407 , i.e. the actual measured data is @xmath364 and @xmath365 satisfying respectively @xmath408 and @xmath409 uniformly in terms of @xmath367 , @xmath368 . 1 .
let @xmath105 be fixed .
the vectors @xmath410 satisfy the formulas @xmath411=-v^t(z^l_m , \theta_i)v^t(z^l_m , \theta_j)+\ ; o\left(a\right)+o(a^{h-1 } ) \delta.\quad\end{aligned}\ ] ] 2 .
let @xmath412 be the constructed fields from step @xmath39 .
we set @xmath413 to be the maximum shift of the centers of the perturbations from step @xmath39 to step @xmath35 , that we denote by @xmath414 and @xmath106 .
let also the minimum distance between these shifted perturbations , i.e. @xmath415 , be of the order @xmath416 , @xmath417 . using the noisy data @xmath375 and
@xmath380 , we have the reconstruction formulas , @xmath106 , + @xmath418\textbf{v}_{\delta}^\top ( \textbf{v}_{\delta}\textbf{v}_{\delta}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}\right]_{_{1,2 } } -\frac{1}{\;4\pi\vert \tilde{z}^2_m-\tilde{z}^1_m\vert}\ ] ] + @xmath419 the formula ( [ true - formula - shifted-2-thsas ] ) will make sense if the error term is of the order @xmath420 as @xmath150 .
it is clear that the noise level should decrease in terms of @xmath26 .
we set @xmath421 and @xmath422 , then this error has the form @xmath423 hence the reconstruction formulas apply if we have the following relations : @xmath424 even if the constant @xmath425 might be quite large . the formulas ( [ v - with - noisy - data - theorem ] ) , to compute the total fields , apply if @xmath426 we observe that the conditions on the noise level to reconstruct the index of refraction ( i.e. using both single and multiple perturbations ) are more restrictive than those needed to construct only the total fields ( i.e. using only single perturbations as it is suggested in the previous literature ) , which is natural of course .
we have shown that by deforming a medium with multiple and close inclusions , we can extract the internal values of the coefficients modeling the medium from the far - field data .
we have tested this idea for the acoustic medium modeled by an index of refraction where the small inclusions are impedance type small scatterers . 1 .
let us consider the more general model @xmath427 , where @xmath428 is the velocity which we assume to be smooth and eventually different from a constant and the small perturbation are also of impedance type . of remark
[ remark ] , this is possible only if the localized perturbations create high contrasts so that , by the leontovich approximation , we replace the contrast by impedance type boundary conditions . ] in this case , arguing as in @xcite , we can derive the same asymptotic expansion as in ( [ x oustdie1 d_m farmain-2])-([fracqcfracmain-2])-([cm - s-1 ] ) . hence proceeding as in the three steps we described above ,
we obtain : @xmath429\textbf{v}^\top(\textbf{v}\textbf{v}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}=\tilde{\textbf{g}}+\;~ o\left ( a^{1-h-2t}+a^{h}\right)\ ] ] and instead of ( [ asymp - g ] ) , we obtain @xmath430+o(\ln(\vert x - z_m\vert)),\ ; \mbox { as } \vert x - z_m\vert \rightarrow 0.\ ] ] from ( [ velocity - c-2 ] ) , we derive @xmath431 + o(\ln(\vert x - z_m\vert)),\ ; \mbox { as } \vert x - z_m\vert \rightarrow 0.\ ] ] combining ( [ velocity - c-3 ] ) and ( [ velocity - c ] ) , with @xmath432 and @xmath433 , we derive the formula @xmath434\textbf{v}^\top(\textbf{v}\textbf{v}^\top)^{-1}\textbf{c}^{-1 } + \textbf{c}^{-1}\right ] _ { 1 , 2}\ ] ] @xmath435 for @xmath436 , recalling that @xmath437 , for @xmath438 .
+ we see that we can derive a direct formula linking the measured data @xmath137 and @xmath138 to the value of the velocity @xmath428 on sampling points @xmath439 in @xmath4 .
the corresponding formulas using the noisy data can also be derived .
2 . the more realistic model would be @xmath427 where the small perturbations occur in the coefficients @xmath428 and @xmath1 and not as small impenetrable obstacles as we dealt with in the present paper .
one needs first to derive the corresponding asymptotic expansion taking into account the parameters modeling these small inclusions and then use them to design a reconstruction method to extract @xmath428 or / and @xmath1 from similar measured data as we did in the present work . in the model we dealt with in this paper , we have chosen the perturbations to be spherical and the surface impedances to be of the form @xmath440 . with this choice , the error in the approximation ( [ x oustdie1 d_m farmain-3 ] ) allows us to recover the fields created at the second order scattering ( and not only the born approximation ) .
this is the key point in order to reconstruct the index of refraction @xmath1 using multiple perturbations . to extend this argument to the case
when the small perturbations occur on the coefficients @xmath428 and @xmath1 , one needs to derive the foldy - lax approximation for such a model allowing to recover fields created at least at the second order ( and , again , not only the born approximation ) .
this issue is related to the full justification of the foldy - lax approximation ( i.e. at higher orders ) .
the justification of the accuracy and the optimal error estimate in the approximation of the scattered field by the foldy - lax field is in its generality a largely open issue but there is an increase of interest to understand it , see for instance .
+ finally , we do believe that this approach can be applied to other multiwave imaging modalities as well .
this work was funded by the deanship of scientific research ( drs ) , king abdulaziz university , under the grant no .
20 - 130 - 36-hici .
the authors , therefore , acknowledge with thanks drs technical and financial support .
hybrid inverse problems and redundant systems of partial differential equations . in _ inverse problems and applications _ , volume 615 of _ contemp .
_ , pages 1547 .
soc . , providence , ri , 2014 . | we deal with the problem of reconstructing material coefficients from the far - fields they generate . by embedding small ( single ) inclusions to these media , located at points @xmath0 in the support of these materials , and measuring the far - fields generated by these deformations we can extract the values of the total field ( or the energies ) generated by these media at the points @xmath0 . the second step is to extract the values of the material coefficients from these internal values of the total field .
the main difficulty in using internal fields is the treatment of their possible zeros . in this work ,
we propose to deform the medium using multiple ( precisely double ) and close inclusions instead of only single ones . by doing so
, we derive from the asymptotic expansions of the far - fields the internal values of the green function , in addition to the internal values of the total fields .
this is possible because of the deformation of the medium with multiple inclusions which generates scattered fields due to the multiple scattering between these inclusions .
then , the values of the index of refraction can be extracted from the singularities of the green function .
hence , we overcome the difficulties arising from the zeros of the internal fields .
we test these arguments for the acoustic scattering by a refractive index in presence of inclusions modeled by the impedance type small obstacles .
* keywords * : inverse acoustic scattering , small inclusions , multiple scattering , refraction index . |
the origin of associated absorption systems ( aas ) in the spectra of qsos with @xmath8 0.01 ( the relative velocity of the aas with respect to the qso , hereafter v , @xmath8 3000 km s@xmath1 ) is not very well understood .
possibilities for their origin include ( i ) interstellar / halo clouds in the host galaxy ( e.g. @xcite ) , ( ii ) material in the core of the active galactic nucleus ( agn ) , within 10 pc of the black hole @xcite , ( iii ) material within 30 kpc of the agn , accelerated by starburst shocks from the inner galaxy ( e.g. * ? ? ?
* ) and ( iv ) clouds in galaxies clustered around the qso ( e.g. * ? ? ?
the study of properties of a large sample of such systems and in particular , their correlation with radio and other properties of the parent qso , should provide clues towards discerning between various scenarios .
several studies of the dependence of the properties of the aas on qso radio properties have been undertaken in the past .
some studies found the frequency of occurrence of aas to depend on the radio properties of the qso ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) other studies failed to find such dependence ( e.g. * ? ? ? * ) but concluded that differences in the results could probably be attributed to various differences in selection of the relatively small samples ( 50 - 100 systems ) .
an excess of aas in radio loud qsos was also found by @xcite in a large sample of sdss qsos . in a previous study ( * ? ? ? * hereafter v08 ) , based on sloan digital sky survey ( sdss ) data release 3 ( dr3 ) , we had studied the average dust extinction and average abundances of a homogeneous sample of 407 aas ( with @xmath8 0.01 ; v @xmath8 3000 km s@xmath1 ) using the method of composite spectra ( york et al .
2006 ; hereafter y06 ) .
definite evidence of dust in the aas was obtained by comparing the composite spectra of the absorber sample with that of a non - absorber ( control ) sample matching in and on a one to one basis .
the dust was found to be of smc type with no evidence for the presence of a 2175 absorption feature .
the amount of dust extinction and the frequency of occurrence of aas in the sample were found to depend on radio properties of the qsos .
a much larger ( by a factor of @xmath9 4 ) sample of systems is now available from the sdss dr7 ( shen & menard 2012 ; hereafter sm12 ) . with this increase in size
, it should be possible to gain further understanding of the associated absorbers .
in particular , we can study the dependence of their dust extinction and star formation rate ( sfr ) ( as measured from the [ o ii]@xmath63727 emission line flux ) , on radio and other properties e.g. the black hole mass ( , determined from the widths of various emission lines ) and eddington ratio ( which is the ratio of bolometric luminosity of the object to its eddington luminosity ) of the qsos .
black hole mass may be indicative of its age . in a merger - driven model of agns ,
supermassive black holes evolve through major mergers which give rise to starbursts and also to accretion onto the nuclear black hole ( sanders et al .
1988 ; hopkins et al . 2005 , 2006 ) . in such models large amounts of gas and dust are funneled inward which fuels the black hole and also causes the obscuration of the young qso .
the dust might be cleared during a transitional phase resulting in the emergence of a luminous blue qso . in this picture ,
reddening is correlated with the evolutionary stage of the qso .
recently , shen & menard ( 2012 ) have shown that qsos with associated absorbers with @xmath8 0.005 ( v @xmath8 1500 km s@xmath1 ) , exhibit enhanced star formation .
they suggest that these absorbers could be large - scale outflows indicative of the transitional phase in a merger - driven evolutionary scenario for qsos . based on the smaller dust extinction and sfr ,
they conclude that the systems with @xmath9 0.005 ( v @xmath9 1500 km s@xmath1 ) originate in intervening absorbers .
in this paper we present the results of our study of the sm12 sample of aas , using the method of composite spectra .
we particularly focus on the dependence of aas properties on the radio and other properties of the qsos .
the details of the sample and sub - samples thereof as well as the method of analysis are presented in section 2 , results are presented in section 3 and conclusions are presented in section 4 .
as mentioned above we used the sample compiled by sm12 ( their table 1 ) from the sdss dr7 .
their sample contains 1937 systems in non - bal qsos with @xmath10 ( v @xmath8 3000 km s@xmath1 ) , spanning the redshift range of 0.4 - 2 and with rest equivalent width ranging from 0.22 to 6.8 .
five of these have @xmath8 0.3 .
there are 92 sight - lines having two aas each and six sight - lines having three aas each . for this study , we focused primarily on qsos having only one aas along their lines of sight , although some statistics for qsos with multiple aas are also presented . also , we restricted our sample to @xmath9 0.3 so that our results can be compared with those of v08 and y06 . our full sample , s1 , thus consists of 1730 aas having @xmath11 0.3 .
we compiled several sub - samples from s1 by dividing it roughly in half , based on various qso and absorber properties : , , radio properties , , and . among the radio properties
, we consider whether the qsos have been radio detected ( rd ) or undetected ( rud ) in the first survey , as well as whether the sources are core dominated ( cd ) or lobe dominated ( ld ) . among our sample of 1730 aas , 263 have been detected while 1341 have been undetected in the first survey .
67 of the 263 rd qsos , 67 are lobe dominated while 196 are core dominated .
the division of s1 based on was done using the criterion of sm12 ( @xmath80 ( v @xmath8 0 km s@xmath1 ) , 0 @xmath0 @xmath8 0.005 ( 0 @xmath0 v @xmath8 1500 km s@xmath1 ) and
@xmath11 0.005 ( v @xmath11 1500 km s@xmath1 ) so that our results can be compared with their conclusions .
the radio and other properties ( e.g. and ) of qsos used for defining the sub - samples were taken from shen et al .
( 2011 ) . throughout this study
, we use the emission redshifts as given by hewett & wild ( 2010 ) and use the relative velocities of aas with respect to these as given by sm12 .
details of the sub - samples are given in table 1 which lists number of systems and the average values of , , , , ( this is given in units of m@xmath5 throughout the paper ) , and .
the method of forming composite spectra is described in detail by y06 and v08 .
we describe it briefly here .
first , the spectra of individual qsos , corrected for galactic reddening , were shifted to the absorber / qso rest - frame and resampled onto a common pixel - to - wavelength scale .
pixels flagged by the spectroscopic pipeline as possibly bad in some way ( stoughton et al . 2002 ) were masked and not used in constructing the composites .
also masked were the pixels within 5 of the expected line positions of detected intervening absorption systems unrelated to the target system .
the geometric mean of all contributing spectra was then calculated for each pixel .
the median / mean composite for [ o ii ] emission line studies was obtained by first fitting the continuum to @xmath12 30 wide regions around 3727 in the qso / absorber rest - frame and then resampling the continuum subtracted spectra to a common wavelength scale .
we calculate the values for various samples by comparing the composite spectrum of each sample with that of the corresponding control samples ( sample of qsos not having aas , matching one to one in and with the qsos in the absorber sample ) , and fitting an smc curve ( pei 1992 ) to the extinction curve so obtained . in deriving this extinction curve ,
we have normalized the two composites at 3000 , which was the value used by sm12 . however , we have verified that normalizing at longer wavelengths does not affect the values .
the values are also independent of whether qso rest - frame or absorber rest - frame composites are used . to construct the absorber rest - frame composite of the control sample , the spectrum of each qso in the sample was shifted to the rest - frame of the absorber towards the corresponding qso ( matching in and ) in the absorber sample .
the typical 1 errors in the derived values are generally smaller than 0.003 ( the errors of the relative flux density values are calculated using the variance formula for the propagation of errors of a geometric mean .
see y06 and v08 for a detailed analysis of errors ) .
the milky way extinction curve does not fit well for any of the samples , and the dust seems to be of smc type , with no evidence of the 2175 bump .
the values so determined are given in the third column of table 2 .
we note that any difference in the values of different sub - samples indicates ( i ) a difference in the dust column density caused either by different gas column density or by different dust - to - gas ratio , or ( ii ) different properties of the dust . in qso absorbers ,
the extinction curve is found to be same ( smc type ) for all the samples .
thus , the dust properties appear to be similar .
the dust content is known to be correlated with ( y06 , v08 , wild et al .
2006 , menard et al .
thus , the dust column may be correlated with and in case we find the values of to be different for sub - samples having similar , then it may indicate a difference in the dust - to - gas ratio for the two sub - samples . for sub - samples s1-s8 and s13-s16 ,
we used the control samples which were used by sm12 ( and kindly provided by them ) . in the construction of these samples
the radio properties of the qsos were not taken into account which is also the case for these sub - samples . however , it is known that the rd qsos are intrinsically redder than the rud qsos ( e.g. see figure 6 of kimball et al . 2011 ; hereafter k11 ) irrespective of presence / absence of aas .
the geometric mean composites for the samples of 4714 rd and 65253 rud dr7 qsos in k11 were kindly provided to us by the authors .
fitting an smc extinction curve to the ratio of the two composites , we estimate the relative between the two samples to be 0.042 .
another selection effect could be important : the sdss qso target selection is made on the basis of ( blue ) color .
however a number of qsos which may not satisfy the color selection criteria are also observed based on other criteria , mainly , their luminosity in other bands like the radio and hence need not have typical qso colors and could be reddened by selection .
about 70 qsos in both s9 and s10 are not color selected , i.e. a much higher fraction of qsos in the rd sample are not color selected and hence could be reddened .
in addition , the intrinsic reddening in rd qsos could also depend on their radio morphology . on the basis of a complete sample of 4714 sdss qsos in dr7 with determined radio properties and with first flux s20 @xmath9 2 mjy
, k11 found that radio sources with unresolved cores have higher reddening .
it is therefore necessary to construct control samples also based on their radio properties for sub - samples s9-s12 .
we constructed such control samples for sub - samples s9 and s10 ( matching one to one in and ) from the rd and rud qsos ( respectively ) in sdss dr7 which do not have aas in their spectra .
the values obtained using these control samples should be independent of any intrinsic reddening in the qsos caused by their radio properties and should reflect the reddening caused by the presence of aas . as the control sample now has the matching radio properties ,
non - color selection should be equally probable in the sub - sample and its control sample .
we also constructed control samples of lobe dominated and core dominated qsos from such qsos in dr7 without aas for s11 and s12 respectively . the values of e(b - v ) for all these sub - samples ( s9-s12 ) , given in table 2 are calculated using these control samples with matching radio properties in addition to matching and .
among dr7 non - bal qsos with redshifts between 0.4 and 2.0 , 68755 qsos have been observed by the first radio survey ( becker , white , and helfand 1995 ) . out of these ,
6366 are radio detected , 4668 are core - dominated , and 1698 are lobe - dominated .
thus , the fraction of rd qsos among all qsos in dr7 in the relevant redshift range is [email protected] . out of 1730 qsos in our sample ( having a single aas each ) ,
1604 have been observed by first : 263 are rd , 196 are core - dominated , and 67 are lobe - dominated .
additionally , out of the 98 qsos in the sm12 sample having multiple aas , 90 have been observed by first : 30 are rd ( 26 being cd ) while 60 are rud qsos .
the fraction of rd qsos among the qsos having aas ( single or multiple ) is [email protected] .
the occurrence of aas is higher by a factor of [email protected] in rd qsos compared to rud qsos .
similarly , while a fraction @[email protected] of rd qsos in our sample have multiple aas , the fraction is @[email protected] for rud qsos in this sample : the incidence of multiple aas is @xmath13 times as likely in rd qsos compared to the rud ones .
these values remain unchanged ( to within 1 ) if the sample is restricted to the systems with @xmath8 0.005 ( v @xmath8 1500 km s@xmath1 , which are intrinsic systems according the results of sm12 .
thus , the dependence of incidence of aas on radio properties is the same for all systems with @xmath80.01 .
the occurrence of aas is significantly related to the radio properties of the qsos which is consistent with the results of most previous studies mentioned in section 1 .
the fraction of core - dominated qsos among the rd dr7 qsos is @[email protected] while in our sample , selected by the presence of ( single or multiple ) aas , it is @[email protected] .
the presence of aas seems to be independent of the morphology of the radio source .
we caution that the numbers here are small and the statistics may not be very meaningful .
these results are unchanged if we restrict the analysis to systems with @xmath8 0.005 . among the 73181 non - bal qsos in the redshift range of 0.4 and 2.0 in dr7 ,
47521 qsos have log ( ) @xmath0 9.1 while 25660 have log ( ) @xmath9 9.1 .
aas are present in a fraction of [email protected] of qsos having smaller black hole mass while for more massive black holes the values are [email protected] .
the values are thus significantly different .
the corresponding values for qsos with log()@xmath0 - 0.81 and @xmath9 - 0.81 are [email protected] and [email protected] respectively .
thus , considering the black hole mass , qsos with older black holes have a higher frequency of occurrence of aas .
no such difference exists for samples of lower and higher eddington ratios .
the reddening for s1 can be directly compared with the reddening obtained for intervening systems by y06 as both samples are from the sdss data and use the same method of composite spectra and as the selection criterion for the two samples are identical except for the range of relative velocities with respect to the qsos .
the for s1 is [email protected] times higher than for the intervening systems for which the is 0.013 .
a comparison with the results of v08 shows that the values obtained here are larger by factors between 1.5 - 2 , while the dependence of on , and radio properties is same .
this could partly be because of the higher fraction ( 15.2%)of rd qsos in our sample as compared to ( 9.9% ) in the sample of v08 and partly a small sample effect . as seen from table 2 ,
the reddening is higher in fainter qsos ( s3 ) compared to brighter ones ( s2 ) .
this could be due to the higher average ( 1.61 ) in s3 compared to that ( 1.32 ) in s2 as reddening is sensitive to the values ( see values for s4 and s5 in table 2 ) .
reddening in aas with @xmath9 0.005 ( s8 ) is significantly smaller than the values for samples with smaller ( s6 and s7 ) as found by sm12 , but is still significantly higher by a factor of 2.1@xmath14 than that in intervening systems .
the sub - sample with higher ( s14 ) is brighter ( see table 1 ) than that with lower ( s13 ) , while the values are equal for both sub - samples .
the sub - sample with lower ( s15 ) is more reddened compared with the sub - sample having higher eddington ratio ( s16 ) .
it is clear from the values that the aas in rd qsos are indeed dustier ( by a factor of [email protected] ) than those in rud qsos . among the average properties ( see table 1 ) of rd and rud qsos ( sub - samples s9 and s10 ) , the rd qsos have only marginally higher , @xmath15 band brightness , and .
the excess extinction in the rd sub - sample over that in the rud qsos thus , can not be possibly be accounted for by the differences in these average properties . as the difference in reddening in rd qsos
can not be accounted by the difference in therefore , as noted in section 2.2 , the aas towards rd qsos may have higher dust - to - gas ratio .
differences can also be seen among the core and lobe - dominated sub - samples , in that core - dominated qsos are 2.0@xmath16 times more reddened .
both classes of rd qsos ( s11 and s12 ) are significantly more reddened compared to the rud qsos .
this is consistent with the results of v08 but is in contrast to the results of k11 who found that only radio sources with unresolved cores have higher reddening , and other classes of rd qsos have reddening comparable to that of the rud qsos . also , the reddening in aas in rud qsos is [email protected] times higher than that in the intervening systems ( y06 ) . to check
if the reddening in rd qsos with aas is correlated with , we divided sub - sample s9 into two roughly equal halves ( s9a and s9b ) depending on .
the results for these are given in tables 2 .
these confirm that the rd qsos with stronger absorption lines are considerably more reddened compared to those having weaker lines ; the difference between the values of the two samples being 0.093 .
the average for the two sub - samples s9a and s9b are 2.5 and 0.9 , which are very similar to those of sub - samples s5 and s4 ( 2.1 and 0.83 ) respectively , for which the differ by only 0.029 .
thus rd qsos have a stronger dependence of reddening on .
we also divided the rud sub - sample ( s10 ) into two halves ( s10a and s10b ) depending on .
the results for these are given in the last rows of tables 2 .
the average values for sub - samples s10a and s10b and the difference in the values for these samples are similar to that between the values for sub - samples s5 and s4 .
the main conclusion from this study is that the aas in rd qsos are significantly more reddened than those in rud qsos which , in turn , are significantly more reddened than the intervening systems .
lcrllrlll s0&sample of sm12&1937&1.43&1.27&1.3&18.56&9.04&-0.64 + s1&full sample&1730&1.46&1.28&2.16&18.59&9.04&-0.63 + s2&[email protected]&862&1.32&1.26&2.46&18.11&9.15&-0.58 + s3&[email protected]&868&1.61&1.30&1.85&19.08&8.93&-0.69 + s4&[email protected] &866&0.83&1.26&2.27&18.46&9.06&-0.64
+ s5&[email protected] &864&2.10&1.30&2.04&18.73&9.01&-0.62 + s6&@xmath21 0.0&510&1.49&1.29&-1.39&18.62&9.05&-0.67 + s7&0.0@xmath0@xmath8 0.005&841&1.47&1.22&1.83&18.64&9.04&-0.70 + s8&@xmath11 0.005&379&1.42&1.40&7.66&18.46&9.01&-0.47 + s9&radio - detected ( rd)&263&1.67&1.18&1.79&18.45&9.03&-0.54 + s10&radio - undetected ( rud)&1341&1.43&1.30&2.25&18.62&9.04&-0.65 + s11&lobe - dominated ( ld)&67&1.46&1.18&1.64&18.45&9.11&-0.42 + s12&core - dominated ( cd)&196&1.74&1.18&1.85&18.45&9.01&-0.58 + s13&log(m@xmath22)@xmath23&874&1.47&1.12&2.23&18.72&8.66&-0.49 + s14&log(m@xmath22)@xmath24&856&1.46&1.44&2.08&18.47&9.42&-0.85 + s15&log(r@xmath25)@xmath0 - 0.81&870&1.55&1.20&1.59&18.78&9.19&-1.09 + s16&log(r@xmath25)@xmath9 - 0.81&860&1.38&1.36&2.73&18.41&8.88&-0.411 + + + cccc s1&full sample&0.041&1.92@xmath26 + s2&[email protected]&0.031&1.84@xmath26 + s3&[email protected]&0.049&2.07@xmath27 + s4&[email protected] &0.026&1.9@xmath27 + s5&[email protected] &0.05-&1.94@xmath26 + s6&@xmath21 0.0&0.047&1.32@xmath27 + s7&0.0@xmath0@xmath8 0.005&0.055&2.32@xmath28 + s8&@xmath11 0.005&0.027&1.52@xmath29 + s9&radio - detected ( rd)&0.097&1.73@xmath26 + s10&radio - undetected ( rud)&0.038&[email protected] + s11&lobe - dominated ( ld)&0.056&[email protected] + s12&core - dominated ( cd)&0.109&[email protected] + s13&log(m@xmath22)@xmath23&0.038&1.76@xmath26 + s14&log(m@xmath22)@xmath24&0.041&2.40@xmath30 + s15&log(r@xmath25)@xmath0 - 0.81&0.058&2.19@xmath27 + s16&log(r@xmath25)@xmath9 - 0.81&0.025&1.55@xmath26 + s9a@xmath31&rd , [email protected]&0.149&[email protected]@xmath32 + s9b@xmath33&rd , [email protected]&0.056&[email protected]@xmath32 + s10a@xmath34&rud , [email protected]&0.057&[email protected]@xmath32 + s10b@xmath35&rud , [email protected]&0.018&[email protected]@xmath32 + + + + + + + + + sm12 found excess emission flux in the [ o ii]@xmath63727 line ( hereafter eefoii ) in the composite spectrum of qsos having aas with @xmath80.005 over that in the composite spectra of the control sample of qsos without aas both made at the emission rest frame . these results were interpreted as indicating that the eefoii originates in the host galaxies of qsos having aas with @xmath80.005 , and that it is a measure of sfr in those galaxies . we discuss the [ o ii ] emission further below . in order to study the emission lines , we constructed continuum - subtracted composites ( median and mean ) of various sub - samples as well as those of the corresponding control samples in the qso rest - frame .
we also constructed corresponding composites in the absorber rest - frames .
for this , the spectrum of each qso in the control sample was shifted to the rest - frame of the absorber towards the corresponding qso ( matching in and ) in the absorber sample .
we observed that the median composite is not very meaningful when constructed in the absorber rest - frame as the peak of the [ o ii]@xmath63727 emission line which gets contribution from the agn and its host galaxy at the emission redshift ) is shifted by different amounts ( depending on the value of ) and as a result the total flux in the line in the median composite is not the same as that in the median composites in the emission rest - frame .
this problem does not appear if we construct the mean composites ( without clipping in the emission line region ) .
we have therefore , used the mean composites for measuring the eefoii .
the values of the ratio of the [ o ii ] flux in the composite spectrum of the sample to that in the composite of the corresponding control sample for various sub - samples using the control samples described in section 2 are listed in column 4 of table 2 . in most cases
these are independent of whether the mean composites are made in the emission or absorption rest - frames , we mention the exception to this below .
it can be seen that a significant eefoii is seen for all sub - samples .
the magnitude of eefoii depends on values of qso brightness , , black hole mass , eddington ratio , and on radio morphology .
the eefoii does not depend on the strength of absorption lines ( s4 and s5 ) .
this is different from what is seen for the intervening systems where the [ o ii ] emission flux is proportional to ( noterdaeme et al .
2010 ; menard et al .
2011 ) . as will be seen below , this is due to the different fractions of rd and rud qsos in these samples .
the sub - samples with lower and higher have lower eefoii . finally , the eefoii in the s7 sample ( 0@xmath0 @xmath80.005 ) is much higher than that in s6 ( @xmath36 ) and s8 ( @xmath110.005 ) .
it thus appears that the hosts of aas with between 0 and 0.005 are different from those with @xmath8 0 and @xmath9 0.005 . in order to confirm that the eefoii arises in the emission rest - frame ,
we have plotted in the left panel of figure 1 , the excess flux in the [ o ii ] line in the absorption rest - frame composite spectra of various dependent sub - samples over that in the composite of the corresponding control samples . for this figure we constructed additional sub - samples having relative velocity between ( i ) 0 and 200 ( = 0 - 0.0007 ) , ( ii ) 200 and 400 ( = 0.0007 - 0.0013 ) , ( iii ) 400 and 600 ( = 0.0013 - 0.002 ) , and ( iv ) 600 and 800 ( = 0.002 - 0.0027 )
. the number of systems in these sub - samples are 214 , 185 , 110 , and 106 respectively .
it is clear that the line centers in the absorber rest - frame composites are progressively shifted towards longer wavelengths with increasing values of the sub - samples .
we have plotted the positions of the [ o ii ] lines at the mean velocities of the samples as vertical lines . for all samples with @xmath90 , these are roughly consistent with the centers of the corresponding emission excess confirming that the [ o ii ] emission is in the qso rest - frame . for @xmath80 sample
the line shows up at @xmath123728 in the absorber rest - frame composite .
the [ o ii ] emission for these systems occurs at the absorber rest - frame and does not show up in the emission rest - frame composite .
it thus seems possible that these absorbers are galaxies which are falling towards the qsos . in the right panel of figure 1 , we have plotted the difference between the composites of the same samples and the composites of the corresponding control samples in the emission rest - frame .
the excess is centered around @xmath123728 and clearly shows that the excess emission occurs at the qso redshift as was observed by sm12 , and may originate in the host galaxy of the individual qsos .
note that the excess emission is not seen here for the @xmath8 0 sample .
we have measured the fwhms of the excess [ o ii ] emission in the composites ( like those plotted in the right panel of figure 1 ) of various sub - samples and these values all fall between 8 and 10 . to estimate the errors in fwhm caused by the errors in
, we generated 100 composite spectra for each of the samples plotted in figure 1 , each time choosing random within the one sigma values given by given by hewett & wild ( 2010 ) .
these errors in are smaller than 0.006 for qsos in our sample and in most cases are in fact smaller than 0.0025 . the one @xmath37 error in fwhms as measured from the fwhms of the of the 100 composites for each of the samples are all smaller than 0.5
. the fwhms , thus , are large and are very similar to the fwhms of the [ o ii ] emission lines themselves in the spectra of the composites of the control samples .
it therefore appears that the excess flux originates in the same regions as the [ o ii ] emission lines of qsos .
figure 6 of k11 shows that the flux of the [ o ii]@xmath63727 line is significantly higher in the geometric mean composite of all radio qsos as compared to that for all rud qsos in dr7 .
k11 have also studied the dependence of emission line fluxes in the median composite spectra for several emission lines on the radio properties of the qsos .
they find that the fluxes depend on the type of radio sources ( see their figure 7 ) . in our samples
the values of eefoii in rd qsos are similar to that in rud qsos ( with aas ) .
the eefoii for absorbers in the lobe - dominated qsos ( s11 ) however , is higher than that for the core - dominated qsos ( s12 )
. thus , on the whole the eefoii in rd and rud qsos is same but in rd qsos the eefoii is dependent on the radio morphology .
it is interesting to note that in cd qsos it is even lower than that in the rud qsos .
a comparison of the eefoii for the rd sub - samples s9a and s9b defined on the basis of , shows that among the rd qsos , the eefoii is much stronger in qsos having aas with larger .
the same also holds if we divide the rud sub - sample s10 into two halves ( s10a and s10b ) depending on .
thus , the eefoii depends on the strength of the lines in the aas in both rd and rud sub - samples .
this dependence on is qualitatively similar to what has been found for intervening systems ( noterdaeme et al .
2010 ; menard et al .
using the largest sample of aas compiled so far , we have found that the incidence of mg ii aas significantly depends on the radio properties of the qsos .
radio detected qsos are [email protected] times more likely to have aas as compared to rud qsos . among the rd qsos
, the incidence of aas does not seem to depend on the morphology of the radio source .
this is inconsistent with earlier studies ( mentioned in section 1 ) that found that the incidence of aas is either independent of the radio properties of the qsos ( vestergaard 2003 ) or it is higher in lobe - dominated qsos ( e.g. aldcroft et al . 1994 ) . we also find a significantly higher frequency of occurrence of aas in qsos with higher black hole masses compared with that in qsos with smaller black hole masses .
it is clear from our results that the radio properties of the qsos play an important role in influencing the reddening properties of the aas .
the dust extinction is higher in the aas in rd qsos than that in rud qsos by a factor of [email protected] . among the rd qsos
, the cd qsos are more reddened by a factor of [email protected] as compared to the ld qsos .
the reddening in the aas in rud qsos is also higher than that in the intervening systems ( by a factor of 2.9@xmath38 ) .
for both types of qsos ( rd and rud ) , the reddening in the aas depends on the strength of absorption lines ; aas with stronger lines show higher reddening . based on the dust extinction alone ,
the aas in both rd and rud qsos appear to be intrinsic to the qso .
the reddening is insensitive to the black hole mass but depends on the eddington ratio in the sense that aas in qsos having lower eddington ratios have higher dust extinction .
the dust extinction is found to be similar in aas with @xmath80 and with between 0 and 0.005 .
the aas with @xmath9 0.005 are found to have smaller dust extinction by a factor of 1.7 - 2.0 , however , the extinction is higher than that in the intervening systems ( y06 ) by a factor of 2.1@xmath14 , and therefore those aas may not be intervening as suggested by sm12 .
we note that there is always a selection effect acting against the very dusty systems which will not be observable in a flux limited survey .
this however , applies both to the aas as well as the intervening systems .
thus , the difference in the dust content of these two types of systems as obtained by comparing our results with those of y06 is real as both are samples are taken from the sdss .
we have earlier shown ( khare et al . 2007 ) , from a study of abundances in samples of dlas and sub - dlas observed at high resolution and also the average abundances in sdss mg ii systems at redshifts similar to what we are studying here , that the obscuration bias is not likely to be important .
either very dusty systems do not exist or there may be a bi - modal distribution of dust and there may exist a population of completely dust obscured qsos .
our results do not apply to such a population .
we have clearly demonstrated that the excess emission in the [ o ii ] line originates at the emission redshift of the qsos ( except for the @xmath80 sample for which it occurs close to absorption redshift ) .
this could be due to star formation in the host galaxy .
however , for all sub - samples , the fwhm of the excess flux distribution is large ( @xmath12 8 - 10 ) and is similar to the fwhm of the total [ o ii ] emission .
thus , the eefoii may also originate in the same regions in the parent agn and the host galaxy which emit the [ o ii ] line and the presence of aas enhances the o ii emission from these regions . in the the whole sample ,
the eefoii does not depend on the mg ii equivalent width because of the different fractions of rd and rud qsos in the dependent sub - samples .
this is clear from the fact that when we consider the rd and rud samples separately , the eefoii is significantly higher for qsos having aas with larger , similar to what is observed for the intervening absorbers .
the values of the eefoii are within the range of fluxes determined for intervening mg ii systems ( noterdaeme et al .
the eefoii is similar for the rd and rud qsos ; however , among the rd qsos , the eefoii is higher for the ld qsos as compared to the cd qsos .
thus , the dust extinction and eefoii seem to be anti - correlated among the cd and ld qsos , which is expected except that the dust extinction seems to be too small to explain the difference in eefoiis .
the eefoii does depend on the and .
eefoii is higher in aas in qsos having higher and having lower .
we have studied the properties of associated absorption systems in the redshift range of 0.4 to 2.0 , in the spectra of 1730 sdss qsos .
the main conclusions are as follows : 1 .
the average dust extinction is found to be of smc type with no evidence for the 2175 bump .
the dust extinction is [email protected] times greater than in the intervening absorbers with similar selection criteria .
by using the control samples comprised only of rd , rud , lobe - dominated and core - dominated qsos , we find that ( a ) the aas in rd qsos are [email protected] times more dusty compared to the aas in rud qsos ; ( b ) the reddening due to aas in rd qsos has a stronger dependence on compared to that in rud qsos ; ( c ) among the rd qsos , the aas in the core - dominated qsos have [email protected] times higher dust extinction compared to those in the lobe - dominated qsos ; ( d ) the reddening in the aas in rud qsos is [email protected] times higher than that in intervening absorbers .
4 . the reddening does not depend on the black hole mass and thus its age .
the reddening does depend on the eddington ratio , systems with smaller have higher reddening .
the occurrence of aas is [email protected] times more likely in rd qsos compared to rud qsos .
the occurrence of multiple aas is [email protected] times more likely in rd qsos than in rud qsos .
8 . among the rd qsos
, the frequency of occurrence of aas appears to be independent of the radio morphology .
the frequency of occurrence of aas depends on black hole mass , qsos with larger have higher rate of incidences .
thus , the incidence rate is higher for older black holes .
the eefoii in the aas samples over that in the control samples originates at the qso redshift , and is consistent with its origin in the qso .
the width of the excess emission is large ( fwhm @xmath12 8 - 10 ) and is similar to the width of the [ o ii ] line itself .
this indicates its origin in the [ o ii ] emitting regions in the agn and its host galaxy .
the presence of aas enhances the o ii emission from the agn and/or the host galaxy .
the eefoii is similar for rd and rud qsos .
12 . for the rd as well as for rud qsos ,
the eefoii depends on , such that the sub - sample with higher has higher eefoii in both cases .
13 . among the rd qsos ,
the eefoii is higher for the ld qsos by a factor of [email protected] compared to the cd qsos .
the eefoii depends on the mass of the black hole and the eddington ratio such that qsos with higher and lower have higher eefoii .
the eefoii and dust extinction in cd and ld qsos are anticorrelated .
the eefoii is similar in magnitude to that found in the intervening absorbers .
based on these results the aas seem to have very different amount of dust and dust - to - gas - ratio as compared to the intervening systems , which seem to depend on the radio properties of the qsos and also on the masses of the central black hole .
the excess [ o ii ] emission occurs at the qso redshift .
the width of the excess emission is similar to that of the emission lines in control samples .
the aas could therefore be intrinsic to the qsos .
even with this large sample of aas , we are possibly able to argue against only two of the possibilities , ( i ) and ( iv ) , mentioned in section 1 for the origin of these systems .
their higher dust content compared to the intervening systems ( even for @xmath9 0.005 systems among the aas ) and its dependence on qso properties argues against their origin in the ism of galaxies clustering around the qso , or in the ism of the host galaxies themselves .
one can argue that the jets from the agn can influence the ism in the host as well as the surrounding galaxies .
however , the higher extinction is seen in both radio loud and radio quiet qsos .
further studies are needed to distinguish between the other two scenarios .
pk thanks csir , india for the grant of emeritus scientist fellowship .
we would like to thank shen and menard for making the absorber sample as well as the control sample available .
we are grateful to the referee for giving detailed comments and suggestions which helped improve the presentation in a significant way .
aldcroft , t. l. , bechtold , j. & elvis , m. 1994 , apjs , 93 , 1 anderson , s. f. , weymann , r. j. , foltz , c. b. & chaffee , f. h. 1987 , aj , 94 , 278 baker , j. c. , hunstead , r. w. , athreya , r. m. , barthel , p. d. , de silva , e. , lehnert , m. d. , & suaners , r. d. e. 2002 , apj , 568 , 592 becker , r. h. , white , r. l. , & helfand , d. j. 1995 , apj , 450 , 559 barlow , t. a. , & sargent , w. l. w. 1997 , aj , 113 , 136 chelouche , d. , menard , b. , bowen , d. v. , & gnat , o. 2008 , apj , 683 , 55 foltz , c. b. , chaffee , f. h. jr . ,
weyman , r. j. , anderson , s. f. 1988 . in qso
absorption lines : probing the universe : proceedings of the qso absorption line meeting , baltimore , md , may 19 - 21 , 1987 .
cambridge and new york , cambridge university press , pp53 - 65 fu , h. , & stockton , a. 2007 , apj , 666 , 794 ganguly , r. , bond , n. a. , charlton , j. c. , eracleous , m. , brandt , w. n. & churchill , c. w. 2001 , apj , 549 , 133 hewett , p. c. & wild , v. 2010 , mnras , 405 , 2302 hopkins , p. f. , hernquist , l. , cox , t. j. , et al .
2005 , apj , 630 , 705 hopkins , p. f. , hernquist , l. , cox , t. j. , et al .
2006 , apjs , 163 , 1 khare , p. , kulkarni , v. p. , peroux , c. et al .
2007 , a&a , 464 , 487 kimball , a. e. , ivezic , z. , witta , p. j. & schneider , d. p. 2011 , aj , 141 , 182 ( k11 ) menard , b. , wild , v. , nestor , d. , quidder , a. , zibetti , s. , rao , s. & turnshek , d. 2011 , mnras , 417 , 801 noterdaeme , p. srianand , r. & mohan , v. 2010 , mnras , 403 , 906 pei y. c. , 1992 , apj , 395 , 130 sanders , d. b. , soifer , b. t. , elias , j. h. , et al .
1988 , apj , 325 , 74 shen , y. , richards , g. t. , strauss , m. a. , et al .
2011 , apjs , 194 , 45 shen , y. & menard , b. 2012 , apj , 748 , 131 ( sm12 ) stoughton c. , lupton , r. h. ; bernardi , m. et al .
, 2002 , aj , 123 , 485 .
vanden berk , d. e. , khare , p. , york , d. g , et al .
2008 , apj , 679 , 239 ( v08 ) vestergaard , m. 2003 , apj , 599 , 116 wild , v. , hewett , p. c. & pettini , m. 2006 , mnras , 367 , 211 wild , v. , kauffmann , g. , white , s. , et al .
2008 , mnras , 388 , 227 york , d. g. , khare , p. , vanden berk , d. e. , et al .
2006 , mnras , 367 , 945 ( y06 ) | we studied dust reddening and [ o ii ] emission in 1730 associated absorption systems ( aas ; relative velocity with respect to qsos , @xmath0 3000 km s@xmath1 ; in units of velocity of light , , @xmath0 0.01 ) with 0.4 @xmath2 2 in the sdss dr7 , focusing on their dependence on the radio and other qso properties .
we used control samples , several with matching radio properties to show ( i ) aas in radio detected ( rd ) qsos cause [email protected] times higher dust extinction than those in radio undetected ( rud ) ones which , in turn , cause [email protected] times the dust extinction in the intervening systems ; ( ii ) aas in core - dominated qsos cause [email protected] times higher dust extinction than in lobe - dominated qsos ; ( iii ) occurrence of aas is [email protected] times more likely in rd qsos than in rud qsos and [email protected] time more likely in qsos having black holes with masses larger than 1.23@xmath4 m@xmath5 than in those with lower mass black holes ; ( iv ) there is excess flux in [ o ii]@xmath63727 emission in the composite spectra of the aas samples compared to those of the control samples , which is at the emission redshift .
presence of aas enhances the o ii emission from the agn and/or the host galaxy .
this excess is similar for both rd and rud samples , and is [email protected] times higher in lobe - dominated compared to core - dominated samples .
the excess depends on the black hole mass and eddington ratio .
all these point to the intrinsic nature of the aas except for the systems with @xmath7 which could be infalling galaxies . |
the noisy signal propagation through an optical fiber is governed by the following normalized stochastic nonlinear schrodinger equation ( snlse ) @xcite @xmath0 where @xmath1 , and @xmath2 is the electrical field propagation in the optical fiber .
the term @xmath3 is a standard circularly symmetric complex white gaussian noise , and @xmath4 is a small scalar .
because of the fiber nonlinearity and chromatic dispersion effect , signals from different users interfere with each other in multiuser communications .
recently , yousefi and kschischang @xcite propose a new transmission scheme , called nonlinear frequency division multiplexing ( nfdm ) , with the help of the nonlinear fourier transform , in which the time domain nonlinear channel can be diagonalized in the ( nonlinear ) spectral domain ( composed by discrete spectrum and continuous spectrum ) into several linear scalar multiplicative channels that are independent of each other .
hence , interference free multiuser communication can be achieved at least in the noise free case , which provides a potential of increasing the data rate of communication . however , modulating continuous spectrum is complicated because a set of integration equations need to be solved . as a result , @xmath5-solitons are of interest in optical fiber communications , which does not have continuous spectrum , and only have @xmath5 discrete eigenvalues on the upper half complex plane .
al @xcite design a special sub - class of @xmath5-solitons , namely spatially periodic signals @xmath2 , as possible input signals for the nfdm , which maintain their shape periodically over @xmath6 . in this paper
, we restrict our input to be ( fundamental ) solitons , where @xmath7 .
although the time domain behavior of solitons with noise is known , the spectral domain analysis is lacking .
our motivation is to study how the noise contaminates the solitons in the spectral domain .
specifically , how are the eigenvalue and the spectral amplitude perturbed due to noise .
this work is of importance because the properties of the noise need to be understood in order to study the fundamental limit of reliably communication @xcite .
we use a split - step nonlinear pulse propagation solver ( nppfs ) @xcite for solving the snlse .
uniformly distributed pseudo random numbers were generated on a computer and then transformed into gaussian noises .
the input of the simulation is an ideal fundamental soliton which follows the form of eq.([eq1 ] ) : @xmath8\right\ } e^{-2j\alpha(z)t - j\left\{arg\left[q(z)\right]+\frac{\pi}{2}\right\ } } , \label{eq2}\ ] ] where @xmath9 and @xmath10 are the real and imaginary part of the eigenvalue and @xmath11 is the spectral amplitude .
we chose @xmath12 , @xmath13 , @xmath14 and @xmath15 for the initial values .
nonlinear curve fittings were applied on the amplitudes and the phases of the output pulses for variables @xmath9 , @xmath10 , @xmath16 and @xmath17 . due to
the existence of background noise in the fft window greatly influences the fitting results , only the range of data where @xmath18 where used .
same calculations were repeated 100 times for each @xmath4 .
the results of the simulations are shown in fig.[fig1 ] . the existence of the noise perturbs both real ( @xmath9 ) and imaginary ( @xmath10 ) part of the eigenvalues , which translates into the changes in the output solitons amplitudes and group velocities . as the noise increases , the fluctuations in both real and imaginary part of the eigenvalues increase .
however , the overall changes are relatively small considering @xmath19 corresponding to approximately 10% noise in the output pulse power .
furthermore , when the noise added is small ( @xmath20 ) , the fluctuation of the real and imaginary part of the eigenvalues are more or less evenly distributed around the mean values . as the noise increases , the distribution of the imaginary part of the eigenvalues shifts towards negative side , which corresponding to a decrease in soliton amplitude .
the reason for the decrease can be understood as the power transfer from the soliton to noise through nonlinear interactions as such cross - phase modulation and modulation instability @xcite .
in this paper , we numerically study the noise contaminates the solitons in the nonlinear fourier domain .
evidences are found that the eigenvalues and the nonlinear spectral amplitudes are effected by the noise .
and the fluctuations in the eigenvalues and nonlinear spectral amplitudes increase with the increase of noise
. however , the absolute change in the eigenvalues and nonlinear spectral amplitudes is negligible and will not influence and implementation of the nonlinear frequency division multiplexing . | in this work , how the noise contaminates signals propagating in an optical fibre is discussed in the nonlinear spectral domain after applying the nonlinear fourier transform .
simulation results about how the soliton parameters in the nonlinear spectral domain perturb is shown . |
the strong field approximation ( sfa ) @xcite is _ the _ underlying theory describing the interaction of intense laser light with atoms @xcite or molecules @xcite .
it allows to calculate photoelectron @xcite or high - harmonic spectra @xcite and provides a deep understanding of the cut - offs and the interferences observed in these spectra .
however , it has been noted that several spectral features are not reproduced by the plain sfa , examples being radial structures at low photoelectron energies @xcite holographic side - lobes @xcite , or the low - energy structure @xcite .
the reason for this failure of the sfa is the neglect of the coulomb force on the outgoing ( or returning ) photoelectron .
attempts to include coulomb effects have a long history @xcite and continue to date @xcite , mainly because more and more advanced detector technology and _ ab initio _ simulations reveal more and more features _ not _ covered by the plain sfa . in general , the more `` differential '' the observable is , the trickier is the proper inclusion of coulomb effects . while the simplest task is to correct total ionization rates @xcite , much more challenging , for instance , is to get the interference pattern in photoelectron momentum spectra right @xcite .
it is well - known that photoelectrons generated by strong - field ionization of atoms have a cut - off energy of @xmath0 if they move directly to the detector ( or @xmath1 if they rescatter once from their parent ion ) ( see , e.g. , @xcite for reviews ) .
these are the cut - offs predicted by the plain sfa ( or the sfa extended for one rescattering event , respectively ) , which are hardly affected by the coulomb correction . on the other hand , electrons emitted from laser - irradiated clusters
may have much higher kinetic energies , especially at resonance @xcite . in this work ,
we apply the methodology developed for the coulomb - corrected sfa based on quantum trajectories @xcite to a sfa that is corrected for the collective electric field in the cluster . in this way we are able to show that it is this collective field , which arises because of the coherent oscillation of the electron cloud with respect to the ionic background , that generates multi-@xmath2 electrons .
this finding confirms the sparc effect ( i.e. , `` surface - plasmon - assisted rescattering in clusters '' ) revealed earlier via classical molecular dynamics simulations @xcite .
the aim of the current paper is to introduce a quantum , sfa - based method for metal clusters that treats the ionization step self - consistently and allows for interference effects .
similar approaches may then be also applied to other situations where collective fields play a role , for instance electron emission from metal nanotips @xcite .
the paper is organized as follows . in section [ sec : theory ] , the basic ingredients , i.e. , the sfa , field - corrected quantum orbits , the rigid sphere model for clusters , and the actual numerical implementation are introduced or reviewed . in section
[ sec : results ] a typical photoelectron spectrum obtained for a na@xmath3-cluster close to resonance is presented , and the origin of the fast electrons is revealed by identifying the relevant , collective - field - corrected quantum orbits . finally , we summarize in section [ sec : concl ] . the equation of motion for the electron sphere in the rigid sphere model is derived in [ app : restoringforce ] .
atomic units are used ( in which , numerically , @xmath4 ) unless otherwise noted .
the collective field by which the quantum orbits of the sfa will be corrected is approximated using the rsm . in the rsm
both electrons and ions are modeled by homogeneously charged spheres .
we assume that the radii and the absolute values of the charges of the electron and the ion sphere are equal .
further , we assume one valence electron per atom ( as in sodium na@xmath5 clusters ) so that the cluster radius reads @xmath6 , with @xmath7 the wigner - seitz radius ( @xmath8 for bulk sodium ) .
both ion and electron number densities are @xmath9 .
driving this system with an external ( laser ) field in dipole approximation @xmath10 leads to the following equation of motion ( cf .
[ app : restoringforce ] for a derivation ) @xmath11 here , @xmath12 is the displacement of the center of the electron sphere with respect to the center of the ion sphere ( located in the origin ) , and = = ^-3/2 is the mie frequency .
equation is the equation of motion for a driven , damped , anharmonic oscillator .
a damping @xmath13 is introduced to prevent the singularity in the excursion amplitude at resonance ( see below ) .
if the laser polarization is linear , e.g. , @xmath14 , the excursion @xmath15 will be along @xmath16 as well . for small excursions of the electron sphere @xmath17 , @xmath18 the harmonic oscillator term @xmath19 will dominate , and the anharmonicities @xmath20 and @xmath21 can be neglected
. then , for an electric field of the form e(t)=_z e_0tthe excursion and phase are @xmath22 excursion amplitude @xmath23 and phase @xmath24 as functions of the wigner - seitz radius @xmath25 are shown in . as expected , for @xmath26 ( i.e. , @xmath27 ) and @xmath28 the free - electron limit @xmath29 , @xmath30 is obtained . in the opposite limit @xmath31 ,
@xmath32 one has @xmath33 , @xmath34 . at resonance , @xmath35 , which occurs for the laser parameters chosen at @xmath36 . for finite laser pulses with an envelope @xmath37
the general analytical solution including transient effects is more involved than . for the purpose of this paper
it is sufficient to apply adiabatically , i.e. , the constant laser amplitude @xmath38 is replaced by @xmath37 .
ion and electron sphere , when centered in the origin , give rise to the same spherical electrostatic potential but with opposite sign , @xmath39 this expression , together with the excursion of the electron sphere @xmath12 in , can be used to construct the cluster potential in which a test electron would move , @xmath40 this potential is shown in and used for the cluster correction in section [ sec : clustercorrection ] . ,
namely the one of the emitted electron .
however , for the acceleration mechanism explored in the current work this effect is negligible ( as long as the net charge of the cluster is not too high ) . ] ( solid line , left axis ) and @xmath24 ( dotted line , right axis ) according as functions of the wigner - seitz radius @xmath25 for @xmath41 , @xmath42 , @xmath43 ( see ) .
the dash - dotted vertical line denotes the @xmath7 where resonance occurs . ]
( solid line ) from at maximum elongation of the electron cloud ( at resonance ) using the parameters given in the caption of .
the dash - dotted lines denote the boundaries of the ion distribution , the dashed lines show the boundaries of the electron cloud .
note the steep part ( i.e. , strong force ) in the region where both spheres overlap . ] within the plain sfa the binding force on the outgoing electron is neglected because the electron is described by the solution of the time - dependent schrdinger equation for a free electron in a ( laser ) field but without binding potential , a so - called gordon - volkov state @xcite @xmath44 here , @xmath45 is the vector potential of the laser field , @xmath46 , @xmath47 is the canonical momentum ( which is related to the electron velocity @xmath48 by @xmath49 ) , and @xmath50 ^ 2\rmd t'\ ] ] is the action .
the gordon - volkov state is given in length gauge ( although expressed in terms of the canonical momentum and the vector potential ) .
the plain sfa transition matrix element in length gauge for the so - called direct electrons ( i.e. , those that move directly to the detector , without hard rescattering ) reads ( see , e.g. , @xcite for reviews ) @xmath51 where @xmath52 ^ 2 + i_{p}\rmd t'\ ] ] is the action including the field - free evolution of the initial bound state @xmath53 with ionization potential @xmath54 . the plain sfa matrix element for given @xmath45 , @xmath55 , and @xmath54 can easily be evaluated numerically .
however , not much insight into _ why _ certain spectral features in @xmath56 appear is gained in that way . in the case of @xmath57 the method of steepest descent can be applied to the time integral in @xcite .
the matrix element is then represented by a sum over all saddle point times @xmath58 @xmath59 here , @xmath60 is a pre - exponential factor that depends on the initial state @xmath55 and is evaluated at the respective saddle point time @xmath58 .
the overall qualitative structure of the photoelectron spectra is not affected by the pre - exponential factor @xmath60 .
in particular , cut - offs are determined by the exponential @xmath61 , not by @xmath60 .
, several @xmath55 ( hydrogenic , gaussians of width @xmath62 , short - range @xmath63-potential - like ) . the results shown in the following are calculated for hydrogenic @xmath55 @xcite . ]
the saddle point times @xmath58 for a given momentum @xmath47 are determined by @xmath64 ^ 2 = -i_p,\ ] ] which can be interpreted as energy conservation at the time instant of ionization . since @xmath47 is real and @xmath54 is positive , the saddle point times @xmath58 are necessarily complex .
every saddle point time represents a so - called quantum trajectory @xmath65 @xcite which starts in the complex time plane at @xmath66 .
typically two saddle point times per cycle contribute to a given @xmath47 .
the propagation obeys newton s equation of motion but in complex space and time .
the plain - sfa quantum trajectories are thus obtained by @xmath67\rmd t ' + \bi{r}(\ts)\,,\ ] ] with initial conditions for @xmath68 still to be chosen .
the plain - sfa matrix element @xmath56 does not depend at all on these initial conditions .
however , the corrected sfa matrix element to be introduced below does . in the atomic case one
may choose @xmath69 ( i.e. , the real part of the electron quantum orbit starts at the position of the nucleus ) .
as a consequence , @xmath68 has to be purely imaginary , @xmath70 .
the other condition is given by the canonical momentum @xmath47 which fixes the initial velocity @xmath71 the electron reaches the classically allowed region at @xmath72 .
the so - called `` tunnel exit '' @xmath73 thus reads @xmath74\rmd t ' + \bi{r}(\ts)\\ & = -\rmi\,\bi{p}\,\mathrm{im}\,\ts + \bi{a}(t_r ) - \bi{a}(\ts ) + \rmi\,\mathrm{im}\,\bi{r}(\ts)\end{aligned}\ ] ] with the excursion of a free electron in a laser field @xmath75 we choose the initial position @xmath76 such that the tunnel exit @xmath73 is real , @xmath77 which leads to a real position ( t ) = _ t_r^t[+(t)]t + ( t_r ) , t t_r for the electron for all real times @xmath78 .
`` after '' the tunnel exit @xcite , e.g. , concerning the analyticity of @xmath65 . ]
the quantum trajectory approach offers a convenient way to correct the sfa for the effect of external potentials on the outgoing electron . without external potential the canonical momentum @xmath47 is conserved in a laser field in dipole approximation .
this fact can be used to recast the action into the form @xmath79\rmd t\ ] ] with the electron velocity @xmath80 and the purely momentum - dependent term @xmath81\rmd t$ ] .
this term factors out of the coherent summation of contributions for a fixed asymptotic momentum @xmath47 since it does not change with the saddle - point time @xmath58 .
therefore it does not contribute to the final ionization probability .
we rewrite the second term in using the gordon - volkov hamiltonian of a free electron in a laser field @xmath82 ^ 2\,$ ] , @xmath83\rmd t\,.\ ] ] the actual correction due to the cluster potential now reads @xmath84 because of this modification the canonical momentum is not conserved anymore .
the corrected velocity becomes _
( t)=_(t)+(t ) with @xmath85 the corrected ( now time - dependent ) canonical momentum . the asymptotic momentum @xmath86 for every @xmath58 needs to be calculated by numerical propagation of the associated trajectory .
the final result for the transition amplitude then reads m_^ = _ m_(^ ( ) ) [ eq : correctedtransitionamplitude ] with m_(^ ( ) ) = f__0 ^ -s_(^ ( ) ) , s_(^())=_^()^t .[eq : correctedtransitionamplitudes ] the actual evaluation of the transition amplitude is done in several steps .
first , the saddle - point equation @xmath87 is solved for randomly or uniformly `` shot '' initial canonical momenta @xmath88 within the momentum range of interest .
we use a complex - root - finding algorithm for finding all @xmath58 for a given @xmath88 . every solution @xmath58 corresponds to one trajectory for which the time integral in @xmath89 needs to be solved .
the latter is splitted according @xmath90 where @xmath91 \rmd t\end{aligned}\ ] ] and @xmath92 \rmd t .
\label{eq : matrixelementtimeintegrals } \ ] ] the cluster correction is neglected during the propagation in complex time , i.e. , in @xmath93 ( during the so - called `` sub - barrier motion '' ) . for a given vector potential @xmath45 this part can thus be solved analytically .
this avoids the solution of newton s equations of motion in complex space and time as we correct the electron trajectory only from the tunnel exit to the detector .
the integral in @xmath94 is over real times only but needs to be computed numerically because of the presence of @xmath95 .
the trajectory for a certain @xmath96 is calculated according to newton s equations of motion in real space and time ( dropping the subscript corr for brevity ) , @xmath97 subject to the initial conditions ( t_r^ ( ) ) = _ , ( t_r^())=(t_r^ ( ) ) - ( ^ ( ) ) , s(t_r^())=0 .
this set of ordinary , coupled differential equations of first order is propagated numerically from @xmath98 to @xmath99 for a sufficiently large @xmath100 such that it is ensured that @xmath101 for @xmath102 does not change significantly anymore .
can be calculated analytically once the laser is off . ] including @xmath103 in the set of differential equations is an efficient way to calculate @xmath94 in . from the results
the individual transition amplitude @xmath104 for every @xmath58 can be calculated . to obtain a full momentum spectrum @xmath105 the steps described above
are carried out for many uniformly or randomly shot initial momenta @xmath106 .
the resulting trajectories are binned according their final momenta @xmath107 . in order to calculate @xmath108 the individual transition amplitudes of the trajectories in a final - momentum bin
are added up coherently . in that way
interference effects are incorporated in @xmath105 .
we consider a sodium cluster consisting of @xmath109 atoms in a three - cycle , @xmath110-shaped laser pulse with a frequency of @xmath42 ( corresponding to @xmath111 nm ) and an amplitude of the electric field of @xmath41 ( corresponding to a laser peak intensity of @xmath112 ) .
the oscillation of the electron cloud with respect to the ions is described by the rsm introduced in .
the only free parameter in this model is the damping factor @xmath13 . in the real cluster system
damping occurs because of electron - ion collisions , landau damping , and emission of electrons from the cluster as a whole . for the simulation we have chosen @xmath113 as small as possible but big enough to ensure that even at resonance electron and ion sphere always overlap . allowing for larger excursions in the rsm while assuming an unaffected shape of the electron sphere
seems unreasonable .
as mentioned already in , the choice of @xmath55 only affects the pre - exponential factor and thus does not change the qualitative features of the photoelectron spectra , in particular the cut - offs .
instead , the ionization potential @xmath54 enters the exponential .
we set it to the experimental value @xmath114 for the ground state of na@xmath3 @xcite .
changes of the ionization potential due to cluster expansion are neglected .
another ingredient in the cluster - sfa is the position of the tunnel exit @xmath73 , which has to be modified to make sure that an emitted electron appears _
outside _ the cluster , as is the case in tunneling ionization . to that end the atomic tunnel exit is shifted outwards by three times the cluster radius . along
the laser polarization direction @xmath16 we thus have @xmath115 .
this procedure seems rather _
ad hoc_. however , the photoelectron spectra are robust with respect to changes in this shift as long as ( i ) the electron starts outside the cluster , i.e. , @xmath116 , and ( ii ) the laser is still on when the above mentioned long sfa - orbits pass through the cluster center .
pump - probe schemes are used in the experiment @xcite to allow the clusters to expand .
the first pulse excites the cluster , the second one drives the ionization . between the pulses , the cluster expands , changing its density and thus the mie frequency .
this behavior is mimicked in the simulations by a variation of the wigner - seitz radius @xmath7 of the cluster .
the pump pulse itself is not simulated .
photoelectron momentum distributions for different wigner - seitz radii @xmath7 are shown in .
the first observation is that the left ( @xmath117 ) and right ( @xmath118 ) cut - off momenta change with @xmath7 in a non - monotonous way .
they are maximum for @xmath119 , i.e. , close to the theoretical value @xmath120 for resonance . hence , not surprisingly , the effect of the rsm - based cluster correction to the sfa is largest at resonance .
for the unexpanded cluster with @xmath121 the spectrum is very similar to the plain - sfa spectrum because the influence of the collective electron motion , modeled by the rsm , is small . for expansions beyond the resonance-@xmath7 the effect of the cluster - correction decreases because the electron sphere excursion decreases .
, a three - cycle , @xmath110-shaped laser pulse with a frequency of @xmath42 and an amplitude of the electric field of @xmath41 . for the unexpanded cluster , @xmath121 ( solid line ) , the mie - frequency is greater than the laser frequency , the resonance occurs for @xmath122 ( dash - dotted line ) , for @xmath123 ( dotted line ) the mie - frequency of the cluster is already smaller than the laser frequency .
@xmath124 trajectories were shot for uniformly distributed @xmath125 in the range @xmath126 . ] but with different prefactors in front of the cluster potential @xmath127 in and . at @xmath128
cluster potential the spectrum is equal to the plain - sfa spectrum . for increasing strength of the potential the cut - off momenta increase , and
a plateau develops . ]
illustrates how the plateaus develop by increasing the strength @xmath129 of the cluster - correction potential , @xmath130 , starting from @xmath131 ( plain sfa ) up to its full value @xmath132 .
the spikes that appear in the plateaus , especially close to the cut - offs , are due to the semi - classical nature of the trajectory method and will be discussed below .
( left axis ) for individual trajectories with @xmath133 , binned in time . the electric field ( dashed )
is included ( right axis ) .
the majority of all trajectories with significant contributions to this part of the momentum spectrum is emitted between @xmath134 and @xmath135 .
these trajectories are selected for further analysis . ] in order to identify the mechanism by which the fast electrons are generated at resonance @xmath119 , we first collect the relevant trajectories with @xmath133 . shows the incoherent sum of the single - trajectory probabilities @xmath136 binned in time .
it is seen that the majority of the trajectories contributing to the plateau is emitted within a certain time window around @xmath137 .
all trajectories starting within this time window have positive final momentum .
in fact , the right plateau in is higher than the left . and @xmath138 , see ) .
the solid lines are the trajectories @xmath139 , the dotted lines are the boundaries of the oscillating electron cloud . ]
( left axis ) , the dotted lines show the combined electric field @xmath140 of laser pulse and cluster potential ( right axis ) acting on the respective trajectory .
the dash - dotted line is the electric field @xmath141 of the laser pulse ( right axis ) . around @xmath142
the trajectories show a strong acceleration caused by the strongly negative total electric field .
comparing with this coincides with the additional interaction of the electrons with the cluster . ]
trajectories with @xmath143 , emission times @xmath144 and one return to the center of the cluster during the propagation are presented in in position space and in in momentum space .
shows the additional interaction with the cluster . in atomic , plain - sfa slang
all these trajectories would be denoted `` long trajectories '' : they pass through the origin without rescattering .
within coulomb - corrected , atomic sfa these trajectories are known to be responsible for the holographic side lobes , for instance @xcite . here , in the cluster - case ,
the are accelerated by the collective cluster field .
this is detailed further in , where , besides the canonical momentum , the total electric field `` seen '' by the electron is plotted for each of the relevant trajectories .
this field consists of the laser field and the field generated by the cluster potential . while the laser is treated in dipole approximation , the cluster potential is strongly position - dependent .
as long as the electron is outside the cluster the total field follows the laser field with only small deviations . upon entering the cluster
the total field suddenly becomes strongly negative , resulting in significant acceleration of the electrons in positive direction .
this is caused by the steep cluster potential in the center at large elongations ( see ) .
after leaving the cluster the total field again closely follows the laser field .
our quantum orbit analysis shows that the acceleration mechanism responsible for the plateau in the spectrum is the additional interaction with the cluster potential in phase with the oscillation of the electron cloud .
the magnitude of the acceleration is directly associated with the maximal elongation of the electron cloud , which limits the total field strength during the electron passage . only electrons arriving at the right phase are accelerated by this process .
this leads to the emission of fast electron bunches within narrow time windows , clearly seen in .
another observation is that the selected group of trajectories has a peak momentum where the trajectories accumulate , resulting in a sharp peak at the cut - off momentum in the photoelectron spectrum .
such caustics are known to appear in semi - classical descriptions of quantum dynamics .
although they are typically smoothed in a full quantum treatment , remnants may survive and be observable @xcite .
we developed a collective - field - corrected strong field approximation ( cfsfa ) that is capable of explaining energetic electron emission from laser - irradiated metal clusters close to resonance .
the approach is basically a combination of the ideas already used for a coulomb - corrected sfa based on quantum orbits and the rigid - sphere model of cluster physics .
the latter is used to estimate the collective field acting on the quantum orbits .
so - called `` long trajectories '' , well known from the quantum orbit analysis of the plain sfa , that revisit the cluster interior with the right phase are accelerated by the collective field .
the advantage of the cfsfa compared to classical calculations is that it treats the ionization step self - consistently .
moreover , the method allows for interference , which is essential for , e.g. , the formation of above - threshold ionization peaks , holographic imaging , or any scheme in which structural information is inferred from interference patterns .
this work was supported by the sfb 652 and project ba 2190/8 of the german science foundation ( dfg ) .
starting from the force on a single charged particle of charge @xmath145 in a potential @xmath146 @xmath147 the force on a charge distribution @xmath148 ( of , e.g. , electrons ) can be written as @xmath149 taking the divergence gives @xmath150 using poisson s equation @xmath151 where @xmath152 is the charge density generating the potential @xmath153 , we obtain the overlap integral @xmath154 the charge densities for the two spheres in the rsm in atomic units read @xmath155 ( electrons ) and @xmath156 ( ions ) , with the heaviside step function @xmath157 and the particle density @xmath158 , which is equal for electrons and ions in the case of a single valence electron per atom .
this reduces the integral to the volume of two spherical caps , which is given by @xmath159 where @xmath160 is the height of each of the two caps .
thus @xmath161 whatever direction of @xmath162 is chosen , the dynamics of the two spheres will stay one - dimensional along this direction .
for that reason the left hand side can be written as @xmath163\end{aligned}\ ] ] with @xmath164 the radial component of @xmath165 . hence @xmath166 as the ion sphere is much heavier than the electron sphere , we place the ion - sphere center in the origin and introduce the displacement of the electron - sphere center @xmath15 . taking into account the vector character of the force @xmath167 , @xmath168 on
the electron cloud of mass @xmath169 leads to @xmath170 which is used in section [ sec : rigidspheremodel ] , with driver @xmath10 and damping @xmath171 added .
10 l. v. keldysh , zh . ksp .
teor . fiz . 47 , 1945 ( 1964 ) ; f. h. m. faisal , j. phys .
b : at . mol .
phys . 6 , l89 ( 1973 ) ; h. r. reiss , phys . rev .
a 22 , 1786 ( 1980 ) .
d. b. miloevi , g. g. paulus , d. bauer and w. becker , j. phys .
b 39 , r203 ( 2006 ) .
d. b. miloevi , phys .
a 74 , 063404 ( 2006 ) .
m. lewenstein , ph .
balcou , m. yu .
ivanov , anne lhuillier , and p. b. corkum , phys .
a 49 , 21172132 ( 1994 ) a. rudenko , k. zrost , c. d. schrter , v. l. b. de jesus , b. feuerstein , r. moshammer , and j. ullrich , j. phys .
b 37 , l407 ( 2004 ) .
d. g. arb , s. yoshida , e. persson , k. i. dimitriou , and j. burgdrfer , phys . rev .
96 , 143003 ( 2006 ) .
r. gopal , k. simeonidis , r. moshammer , t. ergler , m. drr , m. kurka , k .- u .
knel , s. tschuch , c .- d .
schrter , d. bauer , j. ullrich , a. rudenko , o. herrwerth , t. uphues , m. schultze , e. goulielmakis , m. uiberacker , m. lezius , and m. f. kling , phys .
103 , 053001 ( 2009 ) .
y. huismans , a. rouze , a. gijsbertsen , j. h. jungmann , a. s. smolkowska , p. s. w. m. logman , f. lpine , c. cauchy , s. zamith , t. marchenko , j. m. bakker , g. berden , b. redlich , a. f. g. van der meer , h. g. muller , w. vermin , k. j. schafer , m. spanner , m. y. ivanov , o. smirnova , d. bauer , s. v. popruzhenko , and m. j. j. vrakking , science 331 , 61 ( 2011 ) . c. i. blaga , f. catoire , p. colosimo , g. g. paulus , h. g. muller , p. agostini , and l. f. dimauro , nature phys . 5 , 335 ( 2009 )
. w. quan , z. lin , m. wu , h. kang , h. liu , x. liu , j. chen , j. liu , x. t. he , s. g. chen , h. xiong , l. guo , h. xu , y. fu , y. cheng , and z. z. xu , phys .
103 , 093001 ( 2009 ) .
a. m. perelomov and v. s. popov , zh .
52 , 514 ( 1967 ) [ sov .
jetp 25 , 336 ( 1967 ) ] .
v. s. popov , physics uspekhi 47 , 855 ( 2004 ) .
f. h. m. faisal and g. schlegel , j. phys .
b 38 , l223l231 ( 2005 ) .
d. g. arb , j. e. miraglia , m. s. gravielle , k. schiessl , e. persson , and j. burgdrfer , phys . rev . a 77 , 013401 ( 2008 ) .
o. smirnova , m. spanner , and m. ivanov , phys .
a 77 , 033407 ( 2008 ) .
s. v. popruzhenko and d. bauer , j. mod . opt .
55 , 2573 ( 2008 ) . s. v. popruzhenko , g. g. paulus , and d. bauer , phys . rev . a 77 , 053409 ( 2008 ) .
yan , s. v. popruzhenko , m. j. j. vrakking , and d. bauer , phys .
105 , 253002 ( 2010 ) .
m. s. gravielle , d. g. arb , j. e. miraglia , and m. f. ciappina , j. phys .
b 45 , 015601 ( 2012 ) .
l. torlina , o. smirnova , phys .
. a 86 , 043408 ( 2012 ) .
tian - min yan , d. bauer , phys .
a 86 , 053403 ( 2012 ) .
p. mulser and d. bauer , _ high - power laser - matter interaction _
( springer , berlin , 2010 ) , chap . 7 . th .
fennel , t. dppner , j. passig , ch .
schaal , j. tiggesbumker , k .- h .
meiwes - broer , phys .
98 , 143401 ( 2007 ) . t .-
yan , s.v .
popruzhenko , d. bauer , _ trajectory - based coulomb - corrected strong field approximation _ in : _ progress in ultrafast intense laser science ix _ , k. yamanouchi , k. midorikawa ( eds . ) ( springer , heidelberg , 2013 ) .
m. krger , m. schenk , p. hommelhoff , nature 475 , 78 ( 2011 ) .
t. shaaran , m.f .
ciappina , m. lewenstein , phys .
a 87 , 053415 ( 2013 ) .
w. gordon , z. phys .
40 , 117 ( 1926 ) ; d.v .
volkov , ibid .
94 , 250 ( 1935 ) .
w. becker , f. grasbon , r. kopold , d. b. miloevi , g. g. paulus and h. walther , adv . at .
48 , 35 ( 2002 ) .
p. salires , b. carr , l. le droff , f. grasbon , g. g. paulus , h. walther , r. kopold , w. becker , d. b. miloevi , a. sanpera , and m. lewenstein , science 292 , 902 ( 2001 ) . f. lindner , m. g. schtzel , h. walther , a. baltuka , e. goulielmakis , f. krausz , d. b. miloevi , d. bauer , w. becker , and g. g. paulus , phys . rev
95 , 040401 ( 2005 ) .
g.g . paulus and d.
bauer , _ double - slit experiments in the time - domain _ in : _ time in quantum mechanics ii ( lecture notes in physics ) _ ( springer , berlin heidelberg , 2010 ) .
l. torlina , j. kaushal , o. smirnova , phys .
a 88 , 053403 ( 2013 ) .
kappes , m. schr , u. rthlisberger , c. yeretzian , e. schumacher , chem .
143 , 251 ( 1988 ) . | the strong field approximation ( sfa ) formulated in terms of so - called `` quantum orbits '' led to much insight into intense - laser driven ionization dynamics . in plain sfa ,
the emitted electron is treated as a free electron in the laser field alone . however , with improving experimental techniques and more advanced numerical simulations it becomes more and more obvious that the plain sfa misses interesting effects even on a qualitative level .
examples are holographic side lobes , the low - energy structure , radial patterns in photoelectron spectra at low kinetic energies , and strongly rotated angular distributions .
for this reason increasing effort has been recently devoted to coulomb corrections of the sfa . in the current paper , we follow a similar line but consider ionization of metal clusters .
it is known that photoelectrons from clusters can be much more energetic than those emitted from atoms or small molecules , especially if the mie resonance of the expanding cluster is evoked .
we develop a sfa that takes the collective field inside the cluster via the simple rigid - sphere model into account .
our approach is based on field - corrected quantum orbits so that the acceleration process ( or any other spectral feature of interest ) can be investigated in detail . |
the experimental realization of bose - einstein condensation ( bec ) in dilute atomic gases of @xmath0rb @xcite , @xmath1na @xcite , @xmath2li @xcite , @xmath3h @xcite , metastable @xmath4he @xcite , @xmath5rb @xcite , @xmath6k @xcite , @xmath7cs @xcite and @xmath8yb @xcite has stimulated theoretical studies of the properties of ultracold bose gases .
the densities of such systems are typically sufficiently low , that their interactions can be described by a single parameter , the @xmath9-wave scattering length @xmath10 . in addition one
can control not only the strength of these interactions but also whether they are attractive ( @xmath11 e.g. @xmath2li , @xmath5rb , @xmath7cs at low magnetic fields ) or repulsive ( @xmath12 ) using magnetic - field induced feshbach resonances @xcite . whereas a harmonically confined bec with repulsive interactions is stable for any number of atoms , a condensate with attractive interactions is only stable if the atom number @xmath13 is smaller than a critical value @xmath14 @xcite . for @xmath15
the interaction energy exceeds the zero - point kinetic energy and a collapse occurs , as extensively studied in @xcite . in experiments on @xmath2li , the magnetic field is held fixed and the number of condensate atoms grows up to @xmath16 where a partial collapse occurs @xcite . in this system , it should be noted that the attractive interactions can lead to the formation of bright matter - wave soliton trains in elongated optical traps @xcite .
in contrast to the @xmath2li experiments , in the case of @xmath5rb , a feshbach resonance @xcite has been used to switch the scattering length from positive to negative values , producing a condensate with @xmath17 which subsequently collapses @xcite , as modelled in @xcite although much work has been done to study the collapse properties of dilute becs in a single - harmonic trap , an interesting subject is the behaviour of a condensate with attractive interactions in a double - well potential .
magnetic @xcite and optical @xcite double - well potentials have been created in recent experiments and a proposal for a magnetic double - well has been reported @xcite . although condensates with @xmath12 in a double - well system have received considerable theoretical attention @xcite ) , only adhikari @xcite and coullet _ et al . _ @xcite have studied the @xmath11 case . in this paper
, we investigate the stability of a low temperature atomic condensate with attractive interactions in a double - well potential , where the josephson dynamics are induced by the addition of a time - dependent potential gradient .
our analysis is based on the gross - pitaevskii equation .
the aim is to discuss the stability of the condensate in the double - well potential and show how tunnelling between the wells can lead to collapse .
we do not , however , intend to describe the _ collapse dynamics _ , for which , one should at least additionally incorporate a suitable 3-body loss term @xcite .
section 2 introduces the main formalism and briefly reviews the eigenenergies of the double - well potential as a function of the potential gradient in both one and three dimensions .
section 3 discusses the stability of a bec in a three dimensional symmetric double - well as a function of the barrier height .
finally , section 4 considers the time - dependent evolution of a 3d system as the gradient is increased at a constant rate .
we find that a collapse occurs as the gradient is increased above a critical value , which , within our formalism , can be predicted by the eigenstate curves .
at low temperatures , the condensate wavefunction @xmath18 ( normalized to unity ) obeys the dimensionless gross - pitaevskii ( gp ) equation , @xmath19 \psi ( \mbox{\boldmath$r$},t)~. \label{eq : gp3dhou}\end{aligned}\ ] ] in the above equation , @xmath20 is the confining potential and @xmath21 , is the dimensionless effective interaction term , where @xmath22 is the usual three - dimensional scattering amplitude , defined in terms of the _ s_-wave scattering length @xmath10 , the total number @xmath13 of atoms of mass @xmath23 , and the harmonic oscillator length in the transverse direction(s ) @xmath24 , with @xmath25 the corresponding trapping frequency .
the confining potential , @xmath26 describes an axially symmetric potential , with @xmath27 and asymmetry parameter @xmath28 plus a gaussian potential of height @xmath29 , located at @xmath30 .
in addition , a linear potential @xmath31 of gradient @xmath32 is pivoted at the centre of the trap . for @xmath33 ,
the right well has higher potential energy and the trap centre is additionally shifted into the @xmath34 region ; however , this shift is negligible for the parameters studied throughout this work . throughout this paper
we work in dimensionless ( harmonic oscillator ) units , by applying the following scalings : space coordinates transform according to @xmath35 , time @xmath36 , condensate wavefunction @xmath37 and energy @xmath38 ( primes henceforth neglected for convenience ) .
we can also consider the 1d form of eq .
( [ eq : gp3dhou ] ) , which accurately takes into account the transverse dynamics of ` cigar ' condensates ( @xmath39 ) , @xmath40 \psi ( z , t)~ , \label{eq : gp1dhou}\end{aligned}\ ] ] where @xmath41 is normalized to unity , @xmath42 is the 1d self - interaction parameter that matches the 1d and 3d axial density profiles and @xmath43 is the confining potential in the axial direction .
@xmath44 for the double - well for an attractive bose gas as a function of the potential gradient
@xmath45 indicating the self - interaction energy , @xmath46 , and the josephson coupling energy , @xmath47 in each case .
the horizontal dotted grey line corresponds to the zero energy of the two - state model .
the value of the nonlinearity used here is @xmath48 , corresponding to @xmath49 and @xmath50 .
the eigenstates at the centre of the trap are also shown in each case .
we assume a spherical trap geometry ( @xmath51 ) , and a gaussian barrier of height @xmath52 located at the centre of the trap.,width=529 ] the eigenstates of the double - well condensate in 1d and 3d are calculated by substituting @xmath53 , where @xmath54 is the chemical potential ( of the one and three dimensional system respectively ) , and solving the resulting time - independent equation as discussed in @xcite . for small nonlinearities
there are only two levels and , for @xmath55 , the eigenstates are a symmetric ground state @xmath56 and an antisymmetric first excited state , @xmath57 , with equal population in both wells .
sufficiently large interactions lead to the appearance of a loop structure ( see e.g. @xcite ) .
this loop structure appears in the first excited state for @xmath12 , and in the ground state for @xmath11 . in this paper
we consider the latter case ( fig .
1 ) , for which the corresponding wavefunctions at @xmath55 are : ( i ) a symmetric ground state @xmath58 with equal population in both wells , ( ii ) an anti - symmetric state with equal population in both wells and a phase difference of @xmath59 across the trap centre , which we shall henceforth refer to as @xmath60 , and ( iii ) two lower energy , non - symmetric ground state solutions with most of the particles in either the left ( dashed ) or the right ( dotted ) well ( the so - called ` self - trapped ' states @xcite ) . under certain conditions , the eigenenergies @xmath54 can also be reproduced by the two - state model ( see e.g. @xcite ) described by the hamiltonian , @xmath61 where @xmath62 is the fractional population difference , @xmath63 is the potential energy difference between the left ( @xmath64 ) and right ( @xmath65 ) wells , @xmath66 is the nonlinearity and @xmath67 is the josephson coupling energy .
the energy splittings @xmath46 and @xmath47 are indicated in fig . 1 .
for the case of a potential gradient @xmath32 , we can write @xmath68 , where @xmath69 is a numerical factor determined numerically from the gp solution . the picture for the eigenenergy levels as a function of @xmath32 shown in fig . 1 , is only valid in 1d and 3d for a range of nonlinearities .
( dashed lines ) and the josephson energy @xmath70 ( solid ) at @xmath55 as a function of the nonlinearity @xmath71 for the 1d confining trap with @xmath72 .
( b ) typical eigenenergies @xmath54 as a function of the potential gradient @xmath32 for various nonlinearities within regions i - iii . shown
are ( from left to right ) the cases @xmath73 , @xmath74 , @xmath75 and @xmath76 . for @xmath77 , @xmath78 .
the horizontal dotted grey line corresponds to @xmath79.,width=529 ] focusing first on the 1d case , we plot in fig .
2(a ) the two - state model parameters @xmath80 and @xmath47 at @xmath55 ( symmetric double - well ) as a function of @xmath71 .
for @xmath81 , increasing the magnitude of the nonlinearity leads to the appearance of a loop structure in the ground state at a critical point @xmath82 .
as a 1d condensate is stable against collapse @xcite , @xmath80 remains finite and a loop structure is always observed . note that , for a nonlinearity less than a critical value , the splitting @xmath47 becomes negative , signifying an inversion of the lowest two energy eigenvalues .
this can be explained using @xmath83 . although @xmath84 is larger for the first excited state than for the ground state , the first excited state has more negative interaction energy ( as its peak density is higher ) , thereby reducing @xmath54 below the ground state value .
2(b ) shows typical eigenenergy levels for the ground and the first excited states as a function of the potential gradient for different values of the nonlinearity .
the two - state model correctly reproduces the eigenenergy curves for small values of @xmath85 but can not reproduce the inversion of the eigenenergy levels for @xmath86 .
the case @xmath87 has been discussed in our earlier work @xcite . in fig .
3 we consider the 3d case . in fig .
3(a ) we plot the splittings @xmath80 and @xmath47 at @xmath55 as a function of @xmath88 .
in contrast to 1d , for @xmath89 the condensate collapses when the atom number , or magnitude of the nonlinearity , exceeds a critical value .
the collapse appears first in the self - trapping states . in fig .
3(a ) this corresponds to the point where the curve for @xmath80 ( dashed line in fig .
3(a ) ) terminates at the boundary between region i and ii ( indicated by the vertical dotted line in fig .
3(a ) ) . at larger negative nonlinearities ( region i in fig .
3(a ) ) the lowest eigenstates invert ( as in the 1d case ) and at @xmath55 the symmetric states also become unstable at @xmath90 . in fig . 3(b ) we plot typical eigenenergy levels as a function of @xmath32 .
the curves are similar to the 1d case except in the limit of large negative nonlinearities ( region i ) , where a completely different structure is found .
note that region i can not be described by the two - state model . in this region
there is no longer a loop structure as the self trapped states are unstable .
this parameter region is of interest for investigating tunnelling induced collapse , where one begins with a stable symmetric state and adds a potential gradient to induce a collapse in one well .
however , before discussing the dynamical behaviour we consider the stability in the symmetric double - well as a function of the barrier height . and @xmath91 .
( b ) typical eigenenergies shown here are for each of the four regions i - iv in ( a ) , namely ( from left to right ) @xmath92 and @xmath59 . compared to the 1d case ( fig .
2 ) , in 3d there is an additional region ( i ) corresponding to the case where the self - trapped states become unstable .
the horizontal dotted grey line corresponds to @xmath79.,width=529 ]
in this section we study the stability of a bec with attractive interactions in a 3d symmetric double - well trap as a function of the barrier height @xmath29 .
we solve eq .
( [ eq : gp3dhou ] ) numerically using the newton method @xcite . above the critical atom number @xmath14
we no longer find stationary solutions .
a dimensionless constant @xmath93 relating the scattering length @xmath10 with @xmath14 and the properties of the confining trap , is defined by , @xcite @xmath94where @xmath95 , @xmath23 is the mass of the atoms confined in the trap and @xmath96 is the geometrically averaged trap frequency . as a function of @xmath29 for a spherical geometry ( @xmath51 ) .
stability curves of the symmetric @xmath56 ( curve ( ii ) in fig .
1 ) , first excited @xmath57 ( curve ( i ) ) and asymmetric ground states ( iii ) are shown as circles , crosses and squares respectively .
( b ) condensate density @xmath97 along the @xmath98 axis as a function of @xmath29 for the state @xmath56 . shown
are the central @xmath99 ( black ) and peak @xmath100 ( grey ) densities , where @xmath101 is the longitudinal position of maximum density in the double - well configuration ( i.e. centre of each individual well ) . in both figures ,
the plotted lines connect adjacent data points .
the vertical dotted line highlights the critical value @xmath102 , above which the stability curve consists of two branches.,width=453 ] in fig .
4(a ) we plot the critical constant @xmath103 as a function of @xmath29 for the case of a symmetric double - well trap with @xmath51 for the ground and first excited states .
we find that at a critical value of the barrier height , @xmath102 , there are two branches to the stability curve for the ground state .
the upper and lower branches correspond to the symmetric ( fig .
1 , curve ( ii ) ) and asymmetric ( fig .
1 , curve ( iii ) ) eigensolutions respectively , ( plotted by circles and squares in fig .
note also that @xmath93 reaches its maximum value at a height just above that corresponding to the appearance of the second branch .
this maximum can be explained by a minimum in the peak density of the double - well configuration , which is plotted in fig .
4(b ) . as the barrier is raised the condensate splits in two , thus reducing its maximum density .
however , as the trap splits to form two separate condensates , the condensates in each well become compressed and the peak density increases again .
our results remain qualitatively unchanged for different values of @xmath104 .
this picture is similar to that of adhikari @xcite , except for the behaviour at large @xmath29 , where adhikari finds that @xmath93 increases again .
as anticipated , we find that , for @xmath105 , when the system is essentially composed of two separate condensates , the value of @xmath93 in each well tends towards the value in a single harmonic trap containing the same number of atoms as each half of the double - well .
finally , the critical constant @xmath93 as a function of @xmath29 , for the first excited state , @xmath106 ( crosses in fig .
4(a ) ) first increases , reaches a peak value , and then decreases approaching the value for the symmetric ground state for large @xmath29 .
this is expected , as in the limit of large @xmath29 , the density distributions of @xmath107 and @xmath106 become very similar .
the main theme of this section is to investigate the possibility of observing a tunnelling induced collapse . at @xmath108
we prepare a stable condensate in a symmetric double - well , for a value of the nonlinear constant @xmath88 in region i of fig .
in order to induce a collapse , a potential gradient is applied at @xmath109 , i.e. , @xmath110 for @xmath111 , such that the right well has higher potential energy than the left .
subsequently , we study the dynamics leading to the onset of collapse , by solving eq .
( [ eq : gp3dhou ] ) numerically .
note that this qualitative picture should remain correct , even if 3-body loss terms are included in the treatment , although the latter may affect the precise value for the onset of the collapse . as a function of @xmath32 for a system initially prepared in state ( a ) @xmath56 and ( b ) @xmath57 ( black lines ) .
the population difference for the eigenstates are also shown as solid grey lines .
the vertical dashed lines mark the critical gradient at which the eigenstates become unstable ( @xmath112 and @xmath113 for @xmath56 and @xmath57 respectively ) .
the vertical dotted lines describe where the system collapses in the time - dependent simulation .
the collapse occurs when @xmath114 reaches the maximum value @xmath115 ( indicated by horizontal grey lines in ( a ) and ( b ) ) of the number asymmetry predicted by the eigenstates .
( c)-(d ) evolution of the interaction energy @xmath116 ( thick black line ) when the potential gradient @xmath110 is increased at a constant rate @xmath117 for ( c ) @xmath56 and ( d ) @xmath57 , with corresponding eigenenergies shown by grey lines .
other parameters used here : @xmath118 , @xmath91 , @xmath51 and @xmath119.,width=491 ] for attractive interactions , the population difference induced by the addition of the gradient does not follow that of the eigenstate , as shown in fig . 5(a ) .
the effect of the nonlinearity is that the ground state is immediately projected onto a superposition of states .
as the potential gradient is increased , the population in one well reaches a critical value and a collapse occurs .
the critical gradient corresponding to the collapse , shown by the dotted vertical line in fig .
5 , is identified as the point where the interaction energy @xmath120 diverges , see fig .
note that the critical value of the number asymmetry @xmath121 for which the time - dependent collapse occurs is close to the maximum value of @xmath115 found for the eigensolution ( grey horizontal line in fig .
5(a ) ) .
this prediction becomes clearer if we consider a condensate prepared in the first excited state @xmath57 , where the time evolution closely follows the eigensolution , see fig .
5(b ) . in this case
the collapse occurs at exactly the point where the number asymmetry exceeds @xmath115 , see inset of fig .
the critical gradient at which the collapse is observed is found to be essentially independent of the rate @xmath117 at which the gradient is increased .
one can also compare the critical number needed in one well before collapse occurs with the prediction for the symmetric potential shown in fig .
4 . by defining @xmath122 , where @xmath123 is the number of atoms in the well which collapses we find that @xmath124 and 0.467 for ground and excited states in fig . 5 , which is close to the value of @xmath125 predicted by the lower branch of fig .
4 . along the @xmath98-axis @xmath126 as a function of time ( s ) for a bec initially prepared in the @xmath56 ( left ) and @xmath57 ( right ) states with @xmath127 . due to the potential gradient
, tunnelling is induced to the left ( @xmath128 ) well for the @xmath56 state and to the right well for @xmath57 .
the condensate instability occurs at @xmath129 s for @xmath56 and @xmath130 s for @xmath57 .
other parameters as in fig .
5.,width=529 ] finally , we discuss typical experimental parameters required for the demonstration of the tunnelling induced collapse . in the harmonic oscillator units discussed in section 2 , the number of atoms is given by , @xmath131 for @xmath132li atoms and taking @xmath127 and @xmath133 hz , we find @xmath134 which is below the critical value for collapse . for an applied field gradient @xmath135
the collapse occurs at @xmath136 s. the collapse is illustrated by the density plots shown in fig .
we have confirmed that the collapse can be avoided if the potential gradient is ramped up to a value smaller than the critical gradient and then held constant .
this would not hold if the system were very close to the critical region , in which case number fluctuations @xcite could enhance tunnelling and hence induce the collapse at a slightly smaller gradient than that predicted by our simple model .
we have studied the stability of a low temperature atomic bec with attractive interactions in a 1d and 3d double - well potential .
in particular we highlight a regime where the condensate is stable if the population in both wells are approximately equal , but becomes unstable if there is sufficient tunnelling from one well to the other .
we study the dynamics of the system when driven by a time - dependent potential gradient and show that a collapse occurs at a critical gradient predicted by the time - independent solutions . although this picture is expected to be qualitatively correct in low temperature atomic condensates , further work is required to determine the precise details of the onset of collapse and the subsequent collapse dynamics , as well as the effects of fluctuations @xcite and finite temperature @xcite .
we acknowledge funding from the uk epsrc .
99 tiesinga e , verhaar b j , and stoof h t c , phys .
a * 47 * , 4114 ( 1993 ) ; courteille p , freeland r s , heinzen d j , van abeelen f a , and verhaar b j , phys .
lett . * 81 * , 69 ( 1998 ) ; inouye s , andrews m r , stenger j , miesner h -j , stamper - kurn d m and ketterle w , nature * 392 * , 151 ( 1998 ) donley e a , claussen n r , cornish s l , roberts j l , cornell e a , and wieman c e , nature * 412 * , 295 ( 2001 ) ; roberts j l , claussen n r , cornish s l , donley e a , cornell e a , and wieman c e , phys .
lett . * 86 * , 4211 ( 2001 ) | the stability of a low temperature bose - einstein condensate with attractive interactions in one and three dimensional double - well potentials is discussed . in particular , the tunnelling dynamics of a condensate under the influence of a time - dependent potential gradient is investigated .
the condensate is shown to collapse at a critical potential gradient which corresponds to a critical number of atoms in one of the two wells .
the sensitivity of this tunnelling induced collapse could provide a useful tool in the study of condensates with attractive interactions . |
in population dynamics and population genetics a prominent role is played by diffusion processes on @xmath1 or @xmath2^g$ ] driven by stochastic differential equations ( sdes ) of the form @xmath3}}}dt + \sqrt{\sigma^2{{{\bigl(x_t(i)\bigr)}}}}\,db_t(i),\quad i\in g , \end { split}\ ] ] for @xmath4 where @xmath5 is a finite or countable set and @xmath6 , @xmath7 , are independent standard brownian motions and where @xmath8 is a stochastic matrix .
we will refer to the solution @xmath9 of as @xmath10-process .
in appropriate timescales and for suitable choices of @xmath11 and @xmath12 , the component @xmath13 describes the ( rescaled ) _ population size _ on island @xmath7 at time @xmath4 , or the relative frequency of a genetic type that is present on island @xmath7 at time @xmath4 . the linear interaction term on the right - hand side of models a _ mass flow _ between the islands , which might be caused by migration of individuals or a flow of genes .
here we will use a picture from population dynamics .
the coefficient @xmath14 then is the infinitesimal variance of the local population size given its current value @xmath15 .
a classical case is feller s branching diffusions where @xmath16 for @xmath15 .
moreover the drift term @xmath11 describes the growth rate of a local population apart from immigration from other islands and emigration .
a prototypical example is @xmath17 , i.e. logistic growth .
the migration matrix @xmath18 is assumed to be a stochastic matrix . here
we address the question of the maximal effect of the migration matrix @xmath18 for fixed @xmath11 and @xmath12 .
we will show in theorem [ thm : comparison ] that the total mass process of @xmath9 is dominted by the virgin island model , which has been constructed in @xcite .
this dominating process does not depend on @xmath5 or on the migration matrix .
the intuition which leads to a comparison result is as follows .
consider a model with supercritical population - size independent branching and additional deaths due to competition within each island of @xmath5 , i.e. , @xmath11 is a concave function with @xmath19 and @xmath12 is a linear function . now
compare different distributions of individuals over space .
if there is at most one individual on each island , then there are no deaths due to competition until the next birth or migration event . if , however , all individuals are on the same island , then there are deaths due to competition .
the effect of the population - size independent branching is the same in both situations .
we infer that more individuals survive if the distribution of individuals is more uniform over space . now
the migration dynamics which distributes mass uniformly over space would be uniform migration on @xmath5 . as there is no uniform migration on an infinite set @xmath5 ,
we approximate @xmath5 with larger and larger finite subsets and consider uniform migration on the finite subsets .
this intuition leads to the assertion that the total mass of the @xmath10-process is dominated by the total mass of an _ @xmath20-island model _
@xmath21 if we let @xmath22 @xmath23 for @xmath24 the @xmath20-island model @xmath21 is the solution of with @xmath25 and @xmath26 , @xmath27 . we will refer to this @xmath20-island model as @xmath28-process .
the initial configurations @xmath29 are assumed to converge suitably , as @xmath30 , to a sequence @xmath31 whose sum is finite .
our first main result ( theorem [ thm : convergence ] ) establishes convergence of @xmath21 as @xmath30 .
the limiting process turns out to be a _
forest of ( mass ) excursions _ started in @xmath32 , @xmath33 .
this forest of excursions has been constructed and analysed in @xcite and has been denoted as _ virgin island model_. in this model , every migrant populates a new island , similar to the infinite - allele model in which every mutation produces a new allele .
jean bertoin turned this idea into a _ tree of alleles _ @xcite and a _ partition into colonies _
@xcite ( see also section 7 of pardoux and wakolbinger ( 2010 ) for a connection with the virgin island model ) .
in particular , bertoin @xcite shows in case of independent branching that the virgin island model is the diffusion approximation of a discrete mass branching process . in section 2
we will review and discuss the virgin island model .
here is a brief heuristics how a forest of ( mass ) excursions emerges from the @xmath28-processes as @xmath30 .
for finite @xmath20 , every island continuously sends migrants to every other island .
every once a while ( in fact , densely in time ) , a migrant is successful and found offspring which immediately find themselves interacting with the resident population on that island .
for a bounded initial mass , the probability that two `` large '' populations that descended from two different founders coexist at the same time in one and the same island turns out to be negligible as @xmath34 .
thus , in the limit @xmath22 , every emigrant populates a previously unpopulated island .
the evolution of the population size on every freshly populated island is described by a random ( mass ) excursion .
these mass excursions are born ( densely in time ) in a poissonian manner on ever new islands with an intensity proportional to the currently extant mass , and , once born , follow the sde @xmath35 formally , this is described by means of the _ excursion measure _ @xmath36 associated with in the sense of pitman and yor ( 1982 ) ( see also @xcite ) .
the intensity measure with which a path @xmath37 spawns a `` daughter '' excursion born at time @xmath4 is @xmath38 .
the roots of the forest are random paths @xmath39 , @xmath33 , which are independent solutions of with @xmath40 , @xmath41 .
the virgin island model is then a countable familie @xmath42 of islands where island @xmath43 is populated at time @xmath44 and carries mass @xmath45 at time @xmath46 .
one tree of excursions in the forest of excursions is illustrated in figure [ f : excursion_tree ] .
note that figure [ f : excursion_tree ] does not contain the whole tree .
in fact islands are populated by emigrants densely in time but only finitely many excursions started by these emigrants reach a given strictly positive height .
now a noteworthy observation is that the tree structure provides us with independence of disjoint subtrees . putting this differently , the virgin island model is a branching process in discrete time in the sense of jiina ( 1958 ) except that there are now infinitely many types , one for each excursion path . due to this branching structure ,
the virgin island model is easier to study than the @xmath20-island process .
several authors considered analogons of the virgin island model in the case of state - independent branching ; see @xcite for a selection of articles . to state our convergence result of theorem [ thm : convergence ] more formally ,
denote the _ population size spectrum _ for the @xmath28-process as @xmath47 for every @xmath48 .
furthermore define the population size spectrum of the virgin island model as @xmath49 our convergence result assert that the population size spectra converge in distribution , i.e. , @xmath50 in distribution as @xmath30 ; see theorem [ thm : convergence ] for the precise statement .
now we state the comparison result more precisly .
recall that @xmath11 is the infinitesimal mean in a non - spatial situation .
we assume @xmath11 to be _
subadditive_. then a population of size @xmath51 that is separated into two islands experiences ( in sum ) a larger growth rate than a population of the same size that is concentrated on one island . thus the virgin island model should offer in expectation a more prolific evolution of the total mass than a model with the same coefficients @xmath11 and @xmath12 .
the infinitesimal variance @xmath12 has an impact on a comparison in distribution .
more precisely the stochastic order in depends on whether @xmath12 is superadditive , additive or subadditive . in our prototype example of additive @xmath12 ,
the stochastic order is the usual increasing order .
then we have that @xmath52 } } } \leq { { \ensuremath{\mathbbm{e}}}}{{{\bigl[f{{{\bigl(\sum_{(s,\chi)\in{\ensuremath{{{\ensuremath{\mathcal v}}}}}}\chi_{t - s}\bigr)}}}\bigr]}}},\quad t\geq0,\ ] ] for every non - decreasing function @xmath53 . if @xmath12 is superadditve or subadditive , then we will use a concave increasing order and a convex increasing order , respectively .
in fact theorem [ thm : comparison ] is a comparison result not only on the one - dimensional , but on the finite dimensional distributions .
thereby results on the distribution of the total mass of the virgin island model have an immediate impact on the distribution of interacting diffusions with local population regulation . as a very special application of our comparison result
, we obtain a sufficient condition for extinction for interacting diffusions with local population regulation . here
we speak of global extinction if the total mass @xmath54 converges to zero in distribution as @xmath55 whenever @xmath56 .
theorem 2 of @xcite shows that global extinction of the virgin island model is equivalent to @xmath57 as a consequence of , if condition is satisfied , then the @xmath10-process dies out globally no matter what the migration matrix @xmath58 is . here
are further implications of our comparison result .
theorem 2 of @xcite together with implies an upper bound for the survival probability of @xmath9 .
theorem 3 of @xcite yields an upper bound for @xmath59 $ ] if the left - hand side of is strictly smaller than one and an upper bound for the growth rate of @xmath60 $ ] as @xmath55 if fails to hold .
moreover theorem 4 of @xcite implies an upper bound for the growth rate of @xmath61 as @xmath55 if fails to hold .
the limit of the @xmath20-island model is well - known if the random variables @xmath62 are exchangeable with common law @xmath63 , say .
here the situation is different due to the unbounded total mass .
emigrants still choose to migrate to island @xmath64 , say , with rate @xmath65 which is small in the limit @xmath22 .
however , the total mass is of order @xmath20 in the case of a nontrivial exchangeable initial configuration .
consequently there is a continuous immigration of migrants onto island @xmath64 even in the limit @xmath22 .
moreover the dynamics of the @xmath20-island model preserves exchangeability .
therefore the rate of immigration converges to the expected average over all islands as @xmath22 due to de finetti s theorem .
this leads to the following limiting process .
the @xmath20-island process @xmath21 converges in distribution as @xmath22 to @xmath66 where @xmath67 , @xmath68 , are independent processes and solve the mckean - vlasov equation @xmath69}}}\,dt + \sqrt{\sigma^2{{{\bigl(m_t\bigr)}}}}\,db_t,\quad{{\ensuremath{\mathcal l^{}\left(m_0\right)}}}=\nu , \end { split}\ ] ] see e.g. theorem 1.4 in @xcite for the case @xmath70 and @xmath11 bounded and globally lipschitz continuous .
thus the components of the @xmath20-island process become independent in the limit @xmath22 .
kac ( 1957 ) described the first known instance of this phenomenon and called it _ propagation of chaos_. after mckean ( 1967 ) , propagation of chaos has then been intensively studied in the mathematical literature , see sznitman s st .
flour lecture notes @xcite .
our convergence is another instance of propagation of chaos . here
independence in the limiting object is independence of disjoint subtrees . in the prototype example of logistic branching
, the mckean - vlasov equation and the virgin island model can be interpreted as forward and backward process .
corollary [ c : duality ] shows for that case that the solution of the mckean - vlasov equation and the total mass process are dual to each other .
this duality derives from a self - duality of the @xmath20-island model .
this self - duality in turn derives from pathwise forward and backward processes in a graphical representation , see @xcite . so in our prototype example , the virgin island model may be interpreted as backward process of the solution of the mckean - vlasov equation .
here is a selection of models for which one could think of a similar comparison result .
mueller and tribe ( 1994 ) investigate a one - dimensional spde analog of interacting feller branching diffusions with logistic growth .
bolker and pacala ( 1997 ) propose a branching random walk in which the individual mortality rate is increased by a weighted sum of the entire population .
etheridge ( 2004 ) studies two diffusion limits hereof .
the `` stepping stone version of the bolker - pacala model '' is a system of interacting feller branching diffusions with non - local logistic growth .
the `` superprocess version of the bolker - pacala model '' is an analog of this in continuous space .
blath , etheridge and meredith ( 2007 ) study a two - type version hereof , which is a spatial extension of the classical lotka - volterra model . fournier and mlard ( 2004 )
generalize the model of bolker and pacala ( 1997 ) by allowing spatial dependence of all rates . a model in discrete time and discrete space
is constructed in birkner and depperschmidt ( 2007 ) . in that paper
an individual has a poisson number of offspring with mean depending on the current configuration and , once created , offspring take an independent random walk step from the location of their mother .
the virgin island model without immigration has been introduced in @xcite . here
we slightly generalize this model by adding independent immigration of mass .
the virgin island model is an analog of in which every emigrant populates a new island .
islands with positive mass at time zero evolve as the one - dimensional diffusion @xmath71 solving . the following assumption guarantees existence and uniqueness of a strong @xmath72-valued solution of equation , see e.g. theorem iv.3.1 in @xcite .
[ a : a1 ] the set @xmath73 is an interval of length @xmath74 which is either of the form @xmath75 $ ] if @xmath76 or of the form @xmath72 if @xmath77 .
the infinitesimal mean @xmath78 and the infinitesimal variance @xmath79 are locally lipschitz continuous in @xmath73 and satisfy @xmath80 and if @xmath76 , then @xmath81 .
the function @xmath82 is strictly positive on @xmath83 and the function @xmath84 is globally upward lipschitz continuous , that is , @xmath85 whenever @xmath86 where @xmath87 is a finite constant .
furthermore @xmath12 satisfies the growth condition @xmath88 for all @xmath89 where @xmath90 is a finite constant .
note that zero is a trap for @xmath71 , that is , @xmath91 implies @xmath92 for all @xmath44 .
if we need the solution @xmath9 of to be well - defined , then we additionally assume the migration matrix to be substochastic .
[ a : migration ] the set @xmath5 is ( at most ) countable and the matrix @xmath8 is nonnegative and substochastic , i.e. , @xmath93 and @xmath94 for all @xmath95 .
note that assumption [ a : a1 ] together with assumption [ a : migration ] guarantees existence and uniqueness of a solution of with values in @xmath96 .
this follows from proposition 2.1 and inequality ( 48 ) of @xcite by letting @xmath97 for @xmath7 and using monotone convergence .
mass emigrates from each island at rate one and colonizes new islands .
a new population should evolve as the process @xmath71 and should start from a single individual which has mass zero due to the diffusion approximation .
thus we need the law of excursions of @xmath71 from the trap zero .
for this , define the set of excursions from zero by @xmath98 , \,\chi_t=0{\ensuremath{\;\;\forall\;}}t\in(-\infty,0]\cup[t_0,\infty)\bigr\}\ ] ] where @xmath99 is the first hitting time of @xmath100 .
the set @xmath101 is furnished with locally uniform convergence . for existence of the excursion measure @xmath36 and in order to apply the results of @xcite , we need to assume additional properties of @xmath84 and of @xmath82 . for the motivation of these assumptions , we refer the reader to @xcite .
assume @xmath102 for some @xmath103
. then the scale function @xmath104 defined through @xmath105 is well - defined .
[ a : hutzenthaler2009ejp ] the functions @xmath84 and @xmath82 satisfy @xmath106 for some @xmath103 . under assumption [ a : hutzenthaler2009ejp ] , the process @xmath71 hits zero in finite time almost surely and the expected total emigration intensity of the virgin island model is finite , see lemma 9.5 and lemma 9.6 in @xcite . moreover the scale function @xmath107 is well - defined and satisfies @xmath108 . a generic example which satisfies assumption [ a : hutzenthaler2009ejp ] is @xmath109 , @xmath110 where @xmath111 , @xmath112 and @xmath113 . assumption [ a : hutzenthaler2009ejp ] is not met by @xmath114 . assuming [ a : a1 ] and [ a : hutzenthaler2009ejp ] , we now define the excursion measure @xmath36 as the law of @xmath71 started in @xmath115 and rescaled with @xmath116 as @xmath117 .
more formally , it is stated in pitman and yor ( 1982 ) ( see @xcite for a proof ) that there exists a unique @xmath118-finite measure @xmath36 on @xmath101 such that @xmath119}}}=\int f(\chi)q(d\chi)\ ] ] for every bounded continuous @xmath120 for which there exists a @xmath121 such that @xmath122 whenever @xmath123
. the reader might want to think of @xmath36 as describing the evolution of a population founded by a single individual . in the special case @xmath124 , @xmath125 with @xmath126 and @xmath127 , the process @xmath71 is feller s branching diffusion whose law is infinitely divisible .
then the excursion measure coincides with the canonical measure .
next we introduce the state space of the virgin island model .
the construction of the virgin island model is generation by generation .
as @xmath128-th generation , we denote the set of all islands which are non - empty at time zero or which are colonized by individuals immigrating into the system .
all islands being colonized by emigrants from the @xmath128-th generation are denoted as islands of the first generation .
the second generation is colonized from the first generation and so on .
for this generation - wise construction , we use a method to index islands which keeps track of which island has been colonized from which island .
an island is identified with a triple which indicates its mother island , the time of its colonization and the population size on the island as a function of time .
let @xmath129 be the set of all possible islands of the @xmath128-th generation .
for each @xmath130 , define @xmath131 which we will refer to as the set of all possible islands of the @xmath132-th generation .
this notation should be read as follows .
island @xmath133 has been colonized from island @xmath134 at time @xmath135 and carries total mass @xmath45 at time @xmath46 .
denote by @xmath136 the union of all @xmath137 over @xmath138 .
the virgin island model will have values in the set of subset of @xmath136 .
having introduced the excursion measure , we now construct the virgin island model with constant immigration rate @xmath139 and started in @xmath140 .
let @xmath141 be independent solutions of such that @xmath142 almost surely .
moreover let @xmath143 be a poisson point process on @xmath144 with intensity measure @xmath145 } } } = { { \ensuremath{\theta}}}\,dt\otimes q(d\psi).\ ] ] the elements of the poisson point process @xmath146 are the islands whose founders immigrated into the system . next we construct all islands which are colonized from a given mother island .
let @xmath147 be a set of independent poisson point processes on @xmath144 with intensity measure @xmath148 } } } = \chi_{t - s}\,dt\otimes q(d\psi)\quad \iota\in{\ensuremath{{{\ensuremath{\mathcal i}}}}}\cup\{\emptyset\}.\ ] ] the elements of the poisson point process @xmath149 are the islands which descend from the mother island @xmath150 .
all ingredients are assumed to be independent .
now the @xmath128-th generation is @xmath151 the @xmath152-st generation , @xmath153 , is the ( random ) set of all islands which have been colonized from islands of the @xmath132-th generation and the virgin island model @xmath154 is the union of all generations @xmath155 we call @xmath156 the virgin island model with immigration rate @xmath157 and initial configuration @xmath158 .
we begin with convergence of the @xmath20-island process . in this convergence
, we allow the drift function @xmath159 and the diffusion function @xmath160 to depend on @xmath20 in order to include the case of weak immigration .
for example , one could be interested in an @xmath20-island model with logistic branching and weak immigration at rate @xmath161 on each island . in that case , one would set @xmath162 and @xmath163 for @xmath164 .
the equation of the @xmath20-island process now reads as @xmath165}}}dt + \sqrt { \sigma_n^2{{{\bigl(x_t^n(i)\bigr)}}}}\,db_t(i ) \end { split}\ ] ] where @xmath166 and where @xmath167 , @xmath33 , are independent standard brownian motions .
the idea to include weak immigration into a convergence result is due to dawson and greven ( 2010 ) who independently obtain convergence of an @xmath20-island model using different methods .
define @xmath168 for @xmath164 . for the @xmath169-process to converge ,
we need assumptions on @xmath170 and on the initial distribution .
[ a : a1_n_linear ] define @xmath171 .
the functions @xmath172 are locally lipschitz continuous on @xmath73 .
the sequence @xmath173 converges pointwise to @xmath11 as @xmath22 .
in addition , @xmath174 as @xmath22 and @xmath175 for all @xmath48 .
the diffusion functions @xmath176 and @xmath12 are linear , that is , @xmath177 and @xmath178 for some constants @xmath179 .
furthermore @xmath180 converges to @xmath181 as @xmath22 . assumptions [ a : a1 ] and [ a : hutzenthaler2009ejp ] hold for @xmath11 .
moreover @xmath173 is uniformly upward lipschitz continuous in zero , that is , @xmath182 for all @xmath183 , @xmath48 and some constant @xmath87 .
here is an example . if @xmath184 and @xmath185 , then assumption [ a : a1_n_linear ] is satisfied if @xmath186 and @xmath187 .
[ a : initial ] the random variables @xmath188 and @xmath189 are defined on the same probability space for each @xmath48 .
there exists a random permutation @xmath190 of @xmath191 for each @xmath48 such that @xmath192}}}=0.\ ] ] furthermore the total mass of @xmath193 has finite expectation @xmath194 .
if @xmath195 is a summable sequence , then assumption [ a : initial ] is satisfied for @xmath196 and @xmath197 , @xmath198 .
next we introduce the topology for the weak convergence of the @xmath20-island process .
what will be relevant here is not any specific numbering of the islands but the statistics ( or `` spectrum '' ) of their population sizes , described by the sum of dirac measures at each time point , that is , @xmath199 where @xmath200 is the dirac measure on @xmath51 .
the state space of the measure - valued process is the set @xmath201 of finite measures on @xmath73 .
we equip the state space @xmath201 with the vague topology on @xmath202 .
now we formulate the convergence of the @xmath169-process defined in .
[ thm : convergence ] suppose that @xmath203 and @xmath204 satisfy assumption [ a : a1_n_linear ] and that the initial configurations @xmath189 and @xmath188 satisfy assumption [ a : initial ] .
then , for every @xmath205 , we have that @xmath206 in distribution where @xmath156 is the virgin island model with immigration rate @xmath207 and initial configuration @xmath188 . the proof is deferred to section [ sec : convergence_to_vim ] .
[ r : convergence ] for readability we rewrite the convergence in terms of test functions .
the weak convergence is equivalent to @xmath208 } } } = { { \ensuremath{\mathbbm{e}}}}{{{\biggl [ f{{{\biggl(\bigl(\sum_{(\iota , s,\eta)\in{{\ensuremath{\mathcal v } } } } f{{{\bigl(\eta_{t - s}\bigr)}}}\bigr)_{t\leq t}\biggr)}}}\biggr]}}}.\ ] ] for every bounded continuous function @xmath209 on @xmath210,{{\ensuremath{\mathbbm{r}}}}\bigr)}}}$ ] and every continuous function @xmath211 with compact support in @xmath83 . applications often require functions @xmath212 with non - compact support .
the following condition might be useful in that case .
let @xmath213 be a continuous function on @xmath210,{{\ensuremath{\mathbbm{r}}}}\bigr)}}}$ ] satisfying the lipschitz condition @xmath214,{{\ensuremath{\mathbbm{r}}}}\bigr)}}}\ ] ] for some @xmath215 .
in addition let @xmath216 be a continuous function satisfying @xmath217 for all @xmath164 . then following the arguments in the proof of lemma [ l : vanishing_immigration_weak_process ] below
, one can show that holds with @xmath209 and @xmath212 replaced by @xmath213 and @xmath218 , respectively .
the assumptions of theorem [ thm : convergence ] are satisfied for branching diffusions with local population regulation .
a prominent example is the @xmath20-island model with logistic drift @xmath219 and with @xmath220 for @xmath15 and some constants @xmath221 .
more generally , theorem [ thm : convergence ] can be applied if @xmath222 and @xmath220 for @xmath15 where @xmath223 is a concave function with @xmath224 .
we believe that theorem [ thm : convergence ] also holds for non - linear infinitesimal variances such as @xmath225 in case of the wright - fisher diffusion .
our proof requires linearity only for one argument which is the step from equation to equation . in case of logistic branching ,
we obtain a noteworthy duality of the total mass process @xmath226 of the virgin island model with the mean field model @xmath227 defined in . by theorem 3 of @xcite , systems of interacting feller branching diffusions with logistic drift satisfy a duality relation which for the @xmath228-process reads as @xmath229 where the notation @xmath230 refers to the initial configuration @xmath231 and @xmath232 refers to @xmath233 .
this duality is established in @xcite via a generator calculation , in swart ( 2006 ) via dualities between lloyd - sudbury particle models and in @xcite by following ancestral lineages of forward and backward processes in a graphical representation .
now let @xmath22 in .
then the left - hand side converges to the laplace transform of the total mass process of the virgin island model ( without immigration ) and the right - hand side converges to the laplace transform of the mean field model .
this proves the following corollary .
[ c : duality ] let @xmath234 be the total mass process of the virgin island model without immigration starting on only one island .
furthermore let @xmath227 be the solution of , both with coefficients @xmath235 and @xmath124 for @xmath100 where @xmath236 .
then @xmath237 where @xmath238 and @xmath239 refer to @xmath240 and @xmath241 , respectively . together with known results on the mean field model
, this corollary leads to a computable expression for the extinction probability of the virgin island model .
let @xmath234 be as in corollary [ c : duality ] .
then @xmath242 converges to a random variable @xmath243 in distribution as @xmath55 .
if @xmath244 then @xmath245 . if condition fails to hold , then @xmath246 where @xmath247 refers to @xmath248 .
the parameter @xmath249 is the unique solution of @xmath250 and the probability distribution @xmath251 is defined by @xmath252 on @xmath253 where @xmath254 is a normalizing constant .
theorem 2 of @xcite shows convergence in distribution of @xmath242 to @xmath243 as @xmath55 and @xmath245 if holds .
if fails to hold , then corollary [ c : duality ] together with convergence of @xmath242 implies convergence in distribution of @xmath255 to a variable @xmath256 as @xmath55 .
the distribution of @xmath256 is necessarily an invariant distribution of the mean field model and is nontrivial .
lemma 5.1 of @xcite shows that there is exactly one nontrivial invariant distribution for and this distribution is given by .
the second main result is a comparison of systems of locally regulated diffusions with the virgin island model . for its formulation ,
we introduce three stochastic orders which are inspired by cox et al ( 1996 ) .
let @xmath257 and @xmath258 be two stochastic processes with state space @xmath73 .
we say that @xmath259 is dominated by @xmath260 with respect to a set @xmath261 of test functions on path space if @xmath262}}}\leq { { \ensuremath{\mathbbm{e}}}}{{{\bigl [ f({{\ensuremath{\tilde{z}}}})\bigr ] } } } \quad{\ensuremath{\;\;\forall\;}}f\in{{\ensuremath{\mathbbm{f}}}}.\ ] ] the first order is the usual stochastic order @xmath263 in which @xmath259 is dominated by @xmath260 if there is a coupling of @xmath259 and @xmath260 in which @xmath264 is dominated by @xmath265 for all @xmath46 almost surely . assuming path continuity , an equivalent condition is as follows .
denote the set of non - decreasing test functions of @xmath138 arguments by @xmath266 for a set @xmath267 .
furthermore let @xmath268 be the set of non - decreasing functions which depend on finitely many time - space points @xmath269 if there is no space component , then we simply write @xmath270 . in this notation
, @xmath271 is equivalent to @xmath272 , see subsection 4.b.1 in shaked and shanthikumar ( 1994 ) .
we will use two more stochastic orders . in the literature , the set of non - decreasing , convex functions is often used . here
an adequate set is the collection of non - decreasing functions whose second order partial derivatives are nonnegative . as we do not want to assume smoothness , we slightly weaken the latter assumption .
we say for @xmath273 that a function @xmath274 is _
@xmath275-convex _ if @xmath276 note that if @xmath212 is smooth , then this is equivalent to @xmath277 . in addition note that @xmath212 is @xmath278-convex if and only if @xmath212 is convex in the @xmath279-th component .
moreover we say that @xmath212 is @xmath275-concave if @xmath280 is @xmath275-convex . a function
is called directionally convex ( e.g. shaked and shanthikumar 1990 ) if it is @xmath275-convex for all @xmath273 .
such functions are also referred to as l - superadditive functions ( e.g. rueschendorf 1983 ) .
define the set of increasing , directionally convex functions as @xmath281 and similarly @xmath282 with @xmath275-convex replaced by @xmath275-concave. furthermore define @xmath283 and @xmath284 as in with @xmath285 replaced by @xmath286 and @xmath287 , respectively .
now we have introduced three stochastic orders @xmath288 , @xmath289 and @xmath290 .
note that @xmath286 contains all mixed monomials and that @xmath287 contains all functions @xmath291 with @xmath292 .
[ thm : comparison ] assume [ a : a1 ] , [ a : migration ] and [ a : hutzenthaler2009ejp ] .
if @xmath11 is concave and if @xmath12 is superadditive , then we have that @xmath293 if @xmath11 is concave and @xmath12 is subadditive , then inequality holds with @xmath284 replaced by @xmath283 . if @xmath11 is subadditive and @xmath12 is additive , then inequality holds with @xmath284 replaced by @xmath268 . the proof is deferred to section [ sec : comparison_with_the_vim ] .
comparisons of diffusions at fixed time points are well - known .
theorem [ thm : comparison ] provides an inequality for the whole process .
the techniques we develope for this in subsection [ ssec : preservation of convexity ] might allow to generalize the comparison results of bergenthum and rueschendorf ( 2007 ) on semimartingales . the assumption of @xmath11 being subadditive is natural in the following sense .
let us assume that letting two @xmath64-island processes with initial masses @xmath51 and @xmath294 , respectively , evolve independently is better in expectation for the total mass than letting one @xmath64-island process with initial mass @xmath295 evolve .
this assumption implies that @xmath296 for all @xmath297 and thus subadditivity of the infinitesimal mean @xmath11 .
if @xmath12 is not additive , then we need the stronger assumption of @xmath11 being concave for lemma [ l : preservation ] . from theorem [ thm : comparison ] and a global extinction result for the virgin island model
, we obtain a condition for global extinction of systems of locally regulated diffusions . according to theorem 2 of @xcite
, the total mass of the virgin island model converges in distribution to zero as @xmath55 if and only if condition below is satisfied .
together with theorem 2 , this proves the following corollary .
[ cor : global_extinction ] assume [ a : a1 ] , [ a : migration ] and [ a : hutzenthaler2009ejp ] .
suppose that @xmath11 is subadditive and @xmath12 is additive or that @xmath11 is concave and @xmath12 is either superadditive or subadditive .
then @xmath298 implies global extinction of the solution @xmath9 of , that is , @xmath299 as @xmath55 whenever @xmath300 almost surely . in case of logistic branching ( @xmath235 , @xmath124 ) , condition simplifies to condition .
for a second example , let @xmath9 be a stepping stone model with selection and mutation , e.g. @xmath301 $ ] , @xmath302 and @xmath303 where @xmath304 .
it is clear that @xmath9 dies out globally if @xmath305 . corollary [ cor : global_extinction ] now improves this so that @xmath306 implies global extinction .
first we outline the intuition behind the proof .
the virgin island process is a tree of excursions whereas the @xmath20-island process has no tree structure .
it happens in the latter process that different emigrants colonize the same island .
in addition , the @xmath20-island process is not loop - free .
an individual could migrate from island @xmath64 to island @xmath307 and then back to island @xmath64 . that these two effects vanish in the limit as the number of islands tends to infinity
will be established in two separate steps .
the first step ensures that the limit of the @xmath20-island process as @xmath22 is loop - free . for this purpose
, we decompose the @xmath20-island process according to the number of migration steps . throughout the paper
, we say that an individual has _
migration level @xmath308 _ at time @xmath309 if its ancestral lineage contains exactly @xmath310 migration steps .
for example , an individual starting on island @xmath64 at time @xmath128 , moving to island @xmath307 and then back to island @xmath64 has migration level @xmath311 .
let @xmath48 .
we define a system @xmath312 of diffusions such that @xmath313 consists of the mass on island @xmath279 at time @xmath314 with migration level @xmath310 .
recall @xmath315 for all @xmath164 .
define @xmath316 for all @xmath317 and @xmath308 .
let @xmath318 be a solution of @xmath319 for all @xmath309 , @xmath166 and @xmath308 where @xmath320 for @xmath46 and @xmath321 and where @xmath322 is a system of independent standard brownian motions .
here we implicitly used the continuous extension of @xmath323 and of @xmath324 as functions of @xmath325 into the point @xmath326 , where @xmath48 .
any weak solution of will be denoted as a @xmath169-process with migration levels .
see lemma [ l : existence_decomp ] for existence of a weak solution of .
lemma [ l : concentration ] below indicates that the individuals with migration level @xmath310 at a fixed time are concentrated on essentially finitely many islands in the limit @xmath22 .
a later migration event will not hit these essentially finitely many islands because hitting a fixed island has probability @xmath327 .
therefore we expect that all individuals on an island have the same migration level .
inserting this into suggests to consider the solution @xmath328 of @xmath329 where @xmath330 for all @xmath46 and @xmath321 .
we will refer to this solution as the loop - free @xmath169-process or as loop - free @xmath20-island model .
lemma [ l : x_close_to_z ] below establishes the assertion that the distance between the @xmath169-process with migration levels and the loop - free @xmath169-process converges to zero in a suitable sense as @xmath22 .
it turns out that some difficulties arise from the different forms of the diffusion coefficients in the @xmath169-process with migration levels and in the loop - free @xmath169-process .
as we could not resolve these difficulties , we additionally assume for lemma [ l : x_close_to_z ] that @xmath176 is linear .
then we have that @xmath331 for @xmath332 and @xmath333 and the diffusion coefficients in and in are similar . our proof of lemma [ l : x_close_to_z ] is a moment estimate in the spirit of yamada and watanabe ( 1971 ) . in subsection
[ ssec : poisson_limit ] we show that two emigrants colonize different islands in the limit @xmath22 .
let us rephrase this more formally . recall that @xmath334 is independent of @xmath335 .
thus , conditioned on @xmath336 , the loop - free @xmath169-process with migration level @xmath337 evolves as the solution of @xmath338 where @xmath339 , @xmath340 , and @xmath341 .
note that @xmath342 , @xmath321 , are independent and identically distributed .
now we are interested in the total mass @xmath343 as @xmath22 . as @xmath339 and as the immigration rate on island
@xmath279 tends to zero , the process @xmath342 converges to zero as @xmath22 for every @xmath68 .
however , as mass of order @xmath344 immigrates on a fixed island , the probability that the excursion started by these immigrants reaches a certain level @xmath121 , say , is of order @xmath344 as the convergence in indicates .
now as there are @xmath20 independent trials , the poisson limit theorem should imply that @xmath345 where @xmath346 is a poisson point process with intensity measure @xmath347 we will prove in lemma [ l : vanishing_immigration_weak_process ] by reversing time . for convergence of the loop - free @xmath169-process
, we do not need to assume linearity of the diffusion function .
here we may replace assumption [ a : a1_n_linear ] with the following weaker assumption .
[ a : a1_n ] the functions @xmath172 and @xmath348 are locally lipschitz continuous on @xmath73 .
the sequence @xmath349 converges pointwise to @xmath350 as @xmath22 .
in addition , @xmath174 as @xmath22 and @xmath351 for all @xmath48 . the functions @xmath159 and @xmath176 satisfy @xmath352 and if @xmath76 , then @xmath353 .
assumption [ a : a1 ] and [ a : hutzenthaler2009ejp ] hold for @xmath11 and @xmath12 .
moreover @xmath173 is uniformly upward lipschitz continuous in zero , that is @xmath182 for all @xmath183 , @xmath48 and some constant @xmath87 .
the sequence @xmath354 satisfies the uniform growth condition @xmath355 for all @xmath89 , @xmath48 where @xmath90 is a finite constant .
note that if @xmath176 is linear , then assumption [ a : a1_n ] implies assumption [ a : a1_n_linear ] .
some steps of our proof are based on second moment estimates and require the following assumption of uniformly finite second moments of the initial distribution .
this assumption is then relaxed in further steps .
[ a : second_moments ] the initial distribution satisfies that @xmath356}}}<\infty.\ ] ] in this subsection we establish preliminary results such as moment estimates and existence of the processes .
the quick reader might want to skip this subsection .
we begin with weak existence of the @xmath20-island process with migration levels .
[ l : existence_decomp ] assume [ a : a1_n ] .
the @xmath169-process with migration levels exists in the weak sense , that is , equation has a weak solution for every @xmath48 .
as the proof is fairly standard , we only give an outline .
approximate with stochastic differential equations for which weak solutions exist .
for example , approximate @xmath159 and @xmath176 with bounded continuous functions @xmath357 and @xmath358 , respectively .
consider the solution @xmath359 of with @xmath159 and @xmath176 replaced by @xmath357 and @xmath358 , respectively , and which only depends on the migration levels @xmath360 .
then this solution has a weak solution according to theorem v.23.5 and theorem v.20.1 of @xcite as the coefficients are bounded and continuous and the stochastic differential equation is finite - dimensional . show that the formal generator hereof converges to the formal generator associated with .
in addition establish tightness of @xmath361 .
then there exists a converging subsequence and its limit solves the martingale problem associated with , see lemma 4.5.1 in @xcite . from this solution of the martingale problem ,
construct a weak solution of as in theorem v.20.1 of @xcite .
next we prove that the @xmath20-island model with migration levels is indeed a decomposition of the @xmath20-island model .
[ l : decomposition ] assume [ a : a1_n ] .
fix @xmath48 and let @xmath362 be a solution of
. then @xmath363 defined through @xmath364 is the unique solution of the @xmath20-island model corresponding to standard brownian motions defined through @xmath365 for every @xmath321 where @xmath366 .
the process @xmath167 is a continuous martingale with quadratic variation process @xmath367_{t}={{\ensuremath{\delta}}}_{ij}t$ ] .
therefore lvy s
characterization ( e.g. theorem iv.33.1 in @xcite ) implies that defines a standard brownian motion .
moreover it follows from summing over @xmath308 that @xmath368 solves .
pathwise uniqueness of has been established in proposition 2.1 of @xcite . for the following lemmas ,
let @xmath369 be a solution of and let @xmath370 be the solution of . define the stopping times @xmath371 through @xmath372 for every @xmath373 and every @xmath48 .
[ l : first_moment_estimate_x_z ] assume [ a : a1_n ] .
then , for every @xmath374 , there exists a constant @xmath375 such that @xmath376 } } } \leq c_t{{{\bigl(2{{\ensuremath{\theta}}}+\sum_{i=1}^n x_i^n\bigr)}}}\ ] ] for every initial configuration @xmath377 and every @xmath48 . fix @xmath48 and @xmath374 . by assumption [ a : a1_n ]
we have that @xmath378 for all @xmath164 .
according to lemma [ l : decomposition ] , @xmath363 defined through is a solution of .
sum over @xmath321 , stop at time @xmath379 and take expectations to obtain that @xmath380 } } } \leq\sum_{i=1}^n x^n(i ) + \int_0^t l { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n { { \ensuremath{\tilde{x}}}}_{s\wedge{{\ensuremath{\tilde{\tau}}}}_k^n}^n(i)\biggr]}}}+2{{\ensuremath{\theta}}}\,ds\ ] ] for every @xmath381 and @xmath373 .
note that the right - hand side is finite .
now gronwall s inequality implies that @xmath382 } } } \leq { { { \bigl(\sum_{i=1}^n x^n(i)+2{{\ensuremath{\theta}}}t\bigr ) } } } e^{l t}\ ] ] for all @xmath381 and @xmath373 . letting @xmath383 fatou s lemma yields that @xmath384 } } } & = \sup_{t\leq t}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n \liminf_{k\to\infty } { { \ensuremath{\tilde{x}}}}_{t\wedge{{\ensuremath{\tilde{\tau}}}}_k^n}^n(i)\biggr]}}}\\ & \leq \sup_{t\leq t}\liminf_{k\to\infty}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n { { \ensuremath{\tilde{x}}}}_{t\wedge{{\ensuremath{\tilde{\tau}}}}_k^n}^n(i)\biggr ] } } } \leq { { { \bigl(\sum_{i=1}^n x^n(i)+2{{\ensuremath{\theta}}}t\bigr ) } } } e^{l t}. \end{split}\ ] ]
this proves inequality with @xmath385 .
the inequality for the loop - free @xmath20-island process follows similarly .
[ l : essentially_finitely_many_generations ] assume [ a : initial ] and [ a : a1_n ] .
then we have that @xmath386 for every @xmath205 .
we prove inequality for the solution of .
the estimate for the solution of is analogous .
recall @xmath315 for all @xmath164 .
apply it s formula to @xmath387 , take expectations , estimate @xmath388 for all @xmath164 and take suprema to obtain that @xmath389 for all @xmath390 and all @xmath308 .
note that the right - hand side is finite due to lemma [ l : first_moment_estimate_x_z ] and assumption [ a : initial ] .
summing over @xmath391 and applying gronwall s inequality implies that @xmath392 for every @xmath393 .
letting @xmath383 proves .
[ l : second_moment_estimate_x_z ] assume [ a : a1_n ] and [ a : second_moments ] .
then we have that @xmath394}}}<\infty\ ] ] is finite for every @xmath374 .
the analogous assertion holds for the loop - free @xmath169-process .
recall @xmath379 from and fix @xmath48 for the moment . according to lemma [
l : decomposition ] , @xmath395 is an @xmath20-island model .
recall from assumption [ a : a1_n ] that @xmath396 for all @xmath164 .
thus lemma 3.3 of @xcite implies that the @xmath169-process is dominated by the @xmath397-process @xmath398 .
using it s formula and @xmath399 for all @xmath164 , we get that @xmath400 } } } = { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\bigl(\sum_{i=1}^n { { { \ensuremath{\bar{x}}}}_{0}^{n}(i ) } \bigr)^2 \biggr ] } } } } \\ & + \int_0^t 2l_\mu { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\bigl(\sum_{i=1}^n { { { \ensuremath{\bar{x}}}}_{s\wedge{{\ensuremath{\tilde{\tau}}}}_k^n}^{n}(i ) } \bigr)^2 \biggr ] } } } + 4{{\ensuremath{\theta}}}{{\ensuremath{\mathbbm{e}}}}{{{\bigl[\sum_{i=1}^n { { \ensuremath{\bar{x}}}}_{s\wedge{{\ensuremath{\tilde{\tau}}}}_k^n}^n(i)\bigr ] } } } + { { \ensuremath{\mathbbm{e}}}}{{{\bigl[\sum_{i=1}^n \sigma_n^2{{{\bigl({{\ensuremath{\bar{x}}}}_{s\wedge{{\ensuremath{\tilde{\tau}}}}_k^n}^n(i)\bigr)}}}\bigr]}}}\,ds\\ & \leq { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\bigl(\sum_{i=1}^n { { { \ensuremath{\bar{x}}}}_{0}^{n}(i ) } \bigr)^2 \biggr ] } } } + \int_0^t ( 2 l_\mu+l_\sigma ) { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\bigl(\sum_{i=1}^n { { { \ensuremath{\bar{x}}}}_{s\wedge{{\ensuremath{\tilde{\tau}}}}_k^n}^{n}(i ) } \bigr)^2 \biggr ] } } } + ( 4{{\ensuremath{\theta}}}+l_\sigma)c_t\,ds \end{split}\ ] ] for every @xmath381 , every @xmath393 and some constant @xmath401 .
we used lemma [ l : first_moment_estimate_x_z ] for the last inequality .
note that the right - hand side is finite .
applying doob s @xmath402 submartingale inequality ( e.g. theorem ii.70.2 in @xcite ) to the submartingale @xmath403 , using fatou s lemma and applying gronwall s inequality to , we conclude that @xmath404 } } } & \leq 4\sup_{t\leq t } { { \ensuremath{\mathbbm{e}}}}{{{\bigl [ \bigl(\sum_{i=1}^n{{\ensuremath{\bar{x}}}}_t^{n}(i)\bigr)^2\bigr]}}}\\ & = 4\sup_{t\leq t } { { \ensuremath{\mathbbm{e}}}}{{{\bigl [ \bigl(\sum_{i=1}^n { { \ensuremath{{\displaystyle \liminf_{k { \rightarrow}\infty}}}}}{{\ensuremath{\bar{x}}}}_{t\wedge{{\ensuremath{\tilde{\tau}}}}_k^n}^{n}(i)\bigr)^2\bigr]}}}\\ & \leq 4\sup_{t\leq t } { { \ensuremath{{\displaystyle \liminf_{k { \rightarrow}\infty}}}}}\ , { { \ensuremath{\mathbbm{e}}}}{{{\bigl [ \bigl(\sum_{i=1}^n { { \ensuremath{\bar{x}}}}_{t\wedge{{\ensuremath{\tilde{\tau}}}}_k^n}^{n}(i)\bigr)^2\bigr]}}}\\ & \leq 4{{{\bigl[{{{\bigl(4{{\ensuremath{\theta}}}+l_\sigma\bigr)}}}c_t t+{{\ensuremath{\mathbbm{e}}}}\bigl(\sum_{i=1}^n x_0^n(i)\bigr)^2\bigr ] } } } e^{(2l_\mu+l_\sigma)t}. \end{split}\ ] ] the right - hand side is bounded uniformly in @xmath48 due to assumption [ a : second_moments ] .
the proof in the case of the loop - free @xmath20-island model is analogous .
recall @xmath379 from .
next we show that stopping at the time @xmath379 has no impact in the limit @xmath383 .
[ l : tau_theta ] assume [ a : a1_n ] and [ a : second_moments ] .
then any solution of satisfies that @xmath405 for every @xmath205 .
the analogous assertion holds for the loop - free @xmath169-process . rewriting @xmath406
, the assertion follows from the markov inequality and from the second moment estimate of lemma [ l : second_moment_estimate_x_z ] .
next we prove some preliminary results for the solution @xmath407 of .
[ l : second_moment_estimate_y ] assume [ a : a1_n ] .
let @xmath408 be locally square lebesgue integrable function .
then we have that @xmath409 } } } \leq c_t{{{\bigl[x+x^2+\int_s^t\ ! \frac{\zeta_n(r)}{n
} + \bigl(\frac{\zeta_n(r)}{n}\bigr)^2\,dr\bigr]}}}\ ] ] for all @xmath164 , @xmath410 , @xmath48 and some constant @xmath375 which does not depend on @xmath51 , @xmath20 or on @xmath411 .
the proof is similar to the proof of lemma [ l : second_moment_estimate_x_z ] , so we omit it . [
l : y_2_supsum ] assume [ a : a1_n ] .
let @xmath408 be locally square lebesgue integrable function .
furthermore let @xmath412 , @xmath321 , be independent solutions of for every @xmath48 .
then we have that @xmath413 } } } \leq c_t{{{\bigl[\int_s^t\ !
\zeta_n(r ) + \frac{\bigl(\zeta_n(r)\bigr)^2}{n}\,dr\bigr]}}}\ ] ] for all @xmath410 , @xmath48 and some constant @xmath375 which does not depend on @xmath20 or @xmath411 .
the proof is similar to the proof of lemma [ l : second_moment_estimate_x_z ] , so we omit it . [
l : first_moment_estimate ] assume [ a : a1_n ] and fix @xmath205 .
let @xmath414 and @xmath415 be two solutions of with respect to the same brownian motion such that @xmath416 and @xmath417 . if @xmath418\to n{\ensuremath{{\displaystyle \cdot}}}i$ ] are square lebesgue integrable , then for some constant @xmath375 being independent of @xmath419 and @xmath48 we have that @xmath420 for all @xmath48 , for all @xmath421 and all @xmath422 . as in theorem 1 of yamada and watanabe ( 1971 ) @xcite , an approximation of @xmath423 with @xmath424-functions results in @xmath425 taking expectations
, the upward lipschitz continuity of @xmath426 implies that @xmath427 for all @xmath428 .
the right - hand side is finite due to lemma [ l : second_moment_estimate_y ] .
therefore , gronwall s inequality implies with @xmath429 .
now we study the solution @xmath430 of in which @xmath431 .
let @xmath432 denote a fixed point in @xmath83 .
recall the scale function @xmath433 from . define the scale function @xmath434 of @xmath430 through @xmath435 for all @xmath436 .
the next two lemmas involve the speed measures @xmath437 as measures on @xmath73 for every @xmath48 .
note that under assumption [ a : a1_n ] , @xmath438 converges as measure on @xmath83 vaguely to @xmath439 .
[ l : replace_initial_point ] assume [ a : a1_n ] and [ a : hutzenthaler2009ejp ] . if @xmath440 are locally square lebesgue integrable , then @xmath441 for all @xmath442 and all globally lipschitz continuous functions @xmath443 . fix @xmath444 and let @xmath445 be the global lipschitz constant of @xmath212 .
the first moment estimate of lemma [ l : first_moment_estimate ] provides us with the inequality @xmath446 for every @xmath121 .
the last step follows from the vague convergence of @xmath447 to @xmath448 together with dominated convergence and assumption [ a : hutzenthaler2009ejp ] .
the right - hand side of converges to zero as @xmath449 by the dominated convergence theorem together with assumption [ a : hutzenthaler2009ejp ] .
[ l : order_of_m_n ] assume [ a : a1_n ] and [ a : hutzenthaler2009ejp ] .
let @xmath447 be the speed measure of @xmath430 .
if @xmath450 , then @xmath451 for every @xmath121 .
fix @xmath121 and @xmath450 .
integration by parts yields that @xmath452}}}_0^{{\ensuremath{\delta}}}\\ & \quad-\int_0^{{\ensuremath{\delta}}}\frac{-y+{{\ensuremath{\tilde{\mu}}}}_n(y)}{\sigma_n^2(y)/2 } \exp{{{\bigl(\int_{\alpha}^y\frac{c / n}{\sigma_n^2(x)/2}\,dx + \int_0^y\frac{-x+{{\ensuremath{\tilde{\mu}}}}_n(x)}{\sigma_n^2(x)/2}\,dx\bigr)}}}\,dy \end{split}\ ] ] for every @xmath48 .
as @xmath176 is lipschitz continuous in @xmath453 $ ] , @xmath454 .
letting @xmath22 in and applying the dominated convergence theorem shows that @xmath455}}}_0^{{\ensuremath{\delta}}}\end{split}\ ] ] which is equal to one .
we recall the following lemma from @xcite , see lemma 9.8 there .
[ l : finite_excursion_area ] assume [ a : a1 ] and [ a : hutzenthaler2009ejp ] . let @xmath36 be the excursion measure defined through . then @xmath456 the last result of this subsection is a variation of the second moment estimate of lemma [ l : second_moment_estimate_x_z ] .
define the stopping times @xmath457 through @xmath458 for every @xmath373 and every @xmath48 .
[ l : second_moment_estimate_x ] assume [ a : a1_n ] , [ a : hutzenthaler2009ejp ] and [ a : second_moments ] .
then we have that @xmath459}}}<\infty\ ] ] is finite for every @xmath374 and @xmath393 .
lemma 3.3 in @xcite shows that , on the event @xmath460 , @xmath461 is stochastically bounded above by @xmath462 . by assumption [ a : a1_n ]
we have that @xmath463 for all @xmath48 .
together with the second moment estimate of lemma [ l : second_moment_estimate_y ] , this implies that @xmath464}}}\\ & \leq \sum_{i=1}^n { { \ensuremath{\mathbbm{e}}}}{{{\biggl[{{\ensuremath{\mathbbm{e}}}}^{x_0^n(i)}{{{\biggl[\biggl(\sup_{t\leq t } y_{t,0}^{n , k+2\theta}\biggr)^2\biggr]}}}\biggr]}}}\\ & \leq c_t\sum_{i=1}^n { { { \bigl({{\ensuremath{\mathbbm{e}}}}{{{\bigl[x_0^n(i)\bigr]}}}+{{\ensuremath{\mathbbm{e}}}}{{{\bigl[{{{\bigl(x_0^n(i)\bigr)}}}^2\bigr ] } } } + t\frac{k+2\theta}{n}+t\frac{(k+2\theta)^2}{n^2}\bigr)}}}\\ & \leq c_t{{{\bigl(\sup_{m\in{{\ensuremath{\mathbbm{n}}}}}{{\ensuremath{\mathbbm{e}}}}{{\ensuremath{\bigl|x_0^m\bigr| } } } + \sup_{m\in{{\ensuremath{\mathbbm{n}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\bigl[{{\ensuremath{\bigl|x_0^m\bigr|}}}^2\bigr]}}}+t ( k+2\theta)+t(k+2\theta)^2\bigr ) } } } \end{split}\ ] ] for every @xmath48 and some constant @xmath375 .
the right - hand side is finite due to assumptions [ a : hutzenthaler2009ejp ] and [ a : second_moments ] . in this subsection , we prove which is the central step in the proof of theorem [ thm : convergence ] . our proof is based on reversing time in the stationary process . for the time reversal , we consider the following stationary situation .
excursions from zero of the process @xmath71 start at times given by the points of an homogeneous poisson point process on @xmath465 with rate @xmath64 .
this process of immigrating excursions is invariant for the dynamics of @xmath71 restricted to non - extinction . now
the time reversal of an excursion is again governed by the excursion measure , see lemma [ l : time_reversal ] . as
a consequence reversing time in the process of immigrating excursions does not change the distribution .
let us retell the story more formally .
consider a poisson point process @xmath346 on @xmath466 with intensity measure @xmath467 .
then @xmath468 is the process of immigrating excursions .
note that at a fixed time @xmath314 , @xmath469 is a poisson point process on @xmath253 with intensity measure @xmath470 where @xmath448 is the speed measure defined in .
here we used that @xmath471 for @xmath472 $ ] .
the relation between the speed measure and the excursion measure has been established in lemma 9.8 of @xcite by exploiting a well - known explicit formula for @xmath473 .
it is also well - known ( e.g. equation ( 15.5.34 ) in @xcite ) that the speed measure @xmath448 is an invariant measure for the sub - markov semigroup @xmath474 .
this fact can also be seen from by noting that @xmath475 is an entrance measure for this sub - markov semigroup .
thus the process of immigrating excursions is indeed invariant for the dynamics of @xmath71 restricted to non - extinction .
now we show that reversing time in the process of immigrating excursions does not change the distribution of the process .
the process @xmath71 restricted to non - extinction is time - reversible when started in the invariant measure @xmath448 , that is , @xmath476 for every @xmath205 and every non - negative measurable function on @xmath210\bigr)}}}$ ] .
if the speed measure @xmath448 is replaced by the left - hand side of , then can be extended to allow for extinction .
to show this we first formulate the markov property of the excursion measure . definition of @xmath36 as rescaled law of @xmath71 together with the markov property of @xmath71 implies that @xmath477 for all @xmath478 satisfying @xmath479 and every @xmath205 .
[ l : time_reversal ] assume [ a : a1 ] and [ a : hutzenthaler2009ejp ] .
then @xmath480 for all @xmath481 and all measurable functions @xmath482 .
it suffices ( see e.g. theorem 14.12 in @xcite ) to establish for @xmath483 where @xmath484 and @xmath485 . if @xmath486 , then both sides of are infinite . for the rest of the proof , we assume @xmath487 , that is , @xmath488 for at least one @xmath489 .
we may even assume @xmath490 for at least one @xmath489 .
otherwise approximate @xmath491 monotonically from below with test functions which have compact support .
in addition , we may without loss of generality assume @xmath492 .
otherwise use a time translation .
if @xmath493 vanishes on @xmath494 , then is essentially . to see this ,
consider @xmath495}\bigr ) } } } { { \ensuremath{\mathbbm{1}}}}_{y_{t_n}>0}\bigr)}}}m({d}x ) .
\end{split}\ ] ] applying with @xmath496 , reversing the calculation in and substituting @xmath497 shows that @xmath498 we prove with @xmath209 replaced by @xmath493 by induction on @xmath130 .
the base case @xmath499 follows from a time translation .
the induction step @xmath500 follows directly from if @xmath501 . if @xmath502 , then @xmath503 for the second step we used linearity , applied the induction hypothesis and equation and again used linearity . adding and proves the induction step in case of @xmath504 .
the remaining case @xmath505 follows from a similar calculation as in .
this completes the proof of lemma [ l : time_reversal ] .
[ l : speed_excursion_measure ] assume [ a : a1 ] and [ a : hutzenthaler2009ejp ] . then @xmath506}\bigr)}}}m(dx )
= \int \int_{-\infty}^t f{{{\bigl({{{(\eta_{t - t - s})}}}_{t\in[0,t]}\bigr)}}}{d}s\ , q({d}\eta)\ ] ] for all measurable functions @xmath507\bigr)}}}\to[0,\infty)$ ] satisfying @xmath508 and all @xmath390 .
express the speed measure in terms of the excursion measure as in @xmath509}\bigr)}}}m(dx)\\ & = \int_{-\infty}^0 \int { { \ensuremath{\mathbbm{1}}}}_{\eta_{-s}>0}{{\ensuremath{\mathbbm{e}}}}^{\eta_{-s}}f{{{\bigl({{{(y_t)}}}_{t\in[0,t]}\bigr)}}}q(d\eta)\,ds\\ & = \int \int_{-\infty}^\infty{{\ensuremath{\mathbbm{1}}}}_{\eta_{-s}>0 } f{{{\bigl({{{(\eta_{t - s})}}}_{t\in[0,t]}\bigr)}}}{d}s\ , q({d}\eta)\\ & = \int \int_{-\infty}^t f{{{\bigl({{{(\eta_{t - t - s})}}}_{t\in[0,t]}\bigr)}}}q({d}\eta)\ , { d}s . \end{split}\ ] ] the last two steps are the markov property and lemma [ l : time_reversal ] , respectively . with lemma [ l : speed_excursion_measure ] in hand , we now reverse time to prove a first version of the poisson approximation .
[ l : vanishing_immigration ] assume [ a : a1_n ] and [ a : hutzenthaler2009ejp ] . let @xmath510 be a constant .
let @xmath430 be the solution of with @xmath511 and @xmath512 .
then @xmath513 for all @xmath46 and every globally lipschitz continuous function @xmath212 on @xmath72 with @xmath514 .
if @xmath515 , then both sides of are equal to zero .
so for the rest of the proof we assume @xmath450 .
let @xmath121 be arbitrary .
lemma [ l : order_of_m_n ] provides us with @xmath516 as @xmath22 .
thus @xmath517 where @xmath518 is a suitable function of @xmath519 .
the term @xmath518 converges to zero as @xmath449 according to lemma [ l : replace_initial_point ] .
the speed measure @xmath447 is an invariant ( non - probability ) measure for @xmath520 , see e.g. theorem v.54.5 in @xcite .
thus we may reverse time . as we let
@xmath22 , we will exploit that @xmath520 converges weakly to @xmath71 .
in addition , @xmath521 converges vaguely to @xmath522 as @xmath22 as the densities converge .
these observations imply that @xmath523 the last but one step is lemma [ l : speed_excursion_measure ] and the last step follows from the dominated convergence theorem together with lemma [ l : finite_excursion_area ] .
putting and together completes the proof of lemma [ l : vanishing_immigration ] .
next we use induction to generalize lemma [ l : vanishing_immigration ] to test functions which depend on finitely many time coordinates . for this
let @xmath524 be the following set of bounded functions on @xmath210,i\bigr)}}}$ ] which depend on finitely many coordinates and which are globally lipschitz continuous in every coordinate @xmath525 for every @xmath205 .
due to the lipschitz continuity and boundedness of @xmath526 , there exists a constant @xmath527 such that @xmath528,i\bigr)}}}.\ ] ] note that the set @xmath524 is closed under multiplication and separates points .
thus the linear span of @xmath524 is an algebra which separates points .
according to theorem 3.4.5 in @xcite the linear span of @xmath524 is distribution determining and so is @xmath524 .
[ l : vanishing_immigration_general ] assume [ a : a1_n ] and [ a : hutzenthaler2009ejp ] .
suppose that @xmath529\to[0,\infty)$ ] and that @xmath530\to n{\ensuremath{{\displaystyle \cdot}}}i$ ] are square lebesgue integrable and that @xmath531 in @xmath532 as @xmath22 .
let @xmath414 satisfy .
then @xmath533 for all @xmath534 and every function @xmath535 .
let @xmath536 be such that @xmath209 satisfies .
moreover let @xmath209 be bounded by @xmath537 .
lemma [ l : first_moment_estimate ] implies that @xmath538 where @xmath539 is the finite constant from lemma [ l : first_moment_estimate ] .
therefore , it suffices to prove with @xmath540 replaced by @xmath541 .
we begin with the case of @xmath541 being a simple function . w.l.o.g .
we consider @xmath542 where @xmath543 and @xmath544 as we may let @xmath209 depend trivially on further time points .
the proof of is by induction on @xmath132 .
the case @xmath499 has been settled in lemma [ l : vanishing_immigration ] .
for the induction step we split up the left - hand side of into two terms according to whether the process at time @xmath545 is essentially zero or not . in order to formalize the notion `` essentially zero '' ,
let @xmath121 be arbitrary and choose functions @xmath546\bigr)}}}$ ] such that @xmath547 for @xmath548 and @xmath549 for @xmath550 .
furthermore let @xmath539 be the constant from lemma [ l : first_moment_estimate ] and define @xmath551 .
first we consider the case that the process is away from @xmath128 at time @xmath545 .
the following estimate shows that we may discard immigration after time @xmath545 . the moment estimate of lemma [ l : first_moment_estimate ] implies that @xmath552}}}\\ & \leq n{{\ensuremath{\mathbbm{e}}}}^0\phi_{{\ensuremath{\delta}}}{{{\bigl(y_{t_1,s}^{n,\zeta}\bigr)}}}c_f n c_t \frac{1}{n}\max_{i\leq n}c_i t_n { \xrightarrow{n { \rightarrow}\infty}}0 \end{split}\ ] ] for every fixed @xmath121 . here
we used that @xmath553 converges to the zero function in distribution as @xmath22 .
if the process is essentially zero at time @xmath545 ( that is @xmath554 ) , then we may restart the process at time @xmath545 in the state @xmath128 .
the lipschitz continuity of @xmath526 together with the moment estimate of lemma [ l : first_moment_estimate ] provides us with @xmath555}}}\\ & \leq { { \ensuremath{{\displaystyle \limsup_{n { \rightarrow}\infty}}}}}n { { \ensuremath{\mathbbm{e}}}}^0{{{\bigl[{{{\bigl(1-\phi_{{\ensuremath{\delta}}}{{{\bigl(y_{t_1,s}^{n , c_1}\bigr)}}}\bigr)}}}n l_f c_t y_{t_1,s}^{n , c_1}\bigr]}}}\\ & = n l_f c_t c_1 \int_s^{t_1}\int{{{\bigl[{{{\bigl(1-\phi_{{\ensuremath{\delta}}}(\chi_{t_1-r})\bigr)}}}\chi_{t_1-r}\bigr]}}}q(d\chi)\,dr .
\end{split}\ ] ] the last step is lemma [ l : vanishing_immigration ] . by the dominated convergence theorem together with lemma [ l : finite_excursion_area ]
, the right - hand side of converges to zero as @xmath449 .
therefore , using , and applying the induction hypothesis @xmath556}}}\\ & = { { \ensuremath{{\displaystyle \lim_{\delta { \rightarrow}0}}}}}{{\ensuremath{{\displaystyle \lim_{n { \rightarrow}\infty}}}}}n{{\ensuremath{\mathbbm{e}}}}^0{{{\bigl[\phi_{{\ensuremath{\delta}}}{{{\bigl(y_{t_1,s}^{n,\zeta}\bigr ) } } } f_1{{{\bigl(y_{t_1,s}^{n,\zeta}\bigr ) } } } { { \ensuremath{\mathbbm{e}}}}^{y_{t_1,s}^{n,\zeta}}{{\ensuremath{\bar{f}}}}_2{{{\bigl({{{\bigl(y_{t - t_1}\bigr)}}}_{t\in{{\ensuremath{\mathbbm{r}}}}}\bigr)}}}\bigr]}}}\\ & \qquad+{{\ensuremath{{\displaystyle \lim_{\delta { \rightarrow}0}}}}}{{\ensuremath{{\displaystyle \lim_{n { \rightarrow}\infty}}}}}{{\ensuremath{\mathbbm{e}}}}^0{{{\bigl(1-\phi_{{\ensuremath{\delta}}}{{{\bigl(y_{t_1,s}^{n,\zeta}\bigr)}}}\bigr ) } } } f_1(0){{\ensuremath{{\displaystyle \lim_{n { \rightarrow}\infty}}}}}n { { \ensuremath{\mathbbm{e}}}}^0 { { \ensuremath{\bar{f}}}}_2{{{\bigl({{{\bigl(y_{t , t_1}^{n,\zeta}\bigr)}}}_{t\in{{\ensuremath{\mathbbm{r}}}}}\bigr)}}}\\ & = { { \ensuremath{{\displaystyle \lim_{\delta { \rightarrow}0}}}}}\int_s^{t_1}\zeta(r)\int\phi_{{\ensuremath{\delta}}}{{{\bigl(\chi_{t_1-r}\bigr)}}}f_1{{{\bigl(\chi_{t_1-r}\bigr ) } } } { { \ensuremath{\mathbbm{e}}}}^{\chi_{t_1-r}}{{\ensuremath{\bar{f}}}}_2{{{\bigl({{{\bigl(y_{t - t_1}\bigr)}}}_{t\in{{\ensuremath{\mathbbm{r}}}}}\bigr)}}}q(d\chi)\,dr\\ & \qquad+f_1(0)\int_{t_1}^{t}\zeta(r)\int { { \ensuremath{\bar{f}}}}_2{{{\bigl({{{(\chi_{t - r})}}}_{t\in{{\ensuremath{\mathbbm{r}}}}}\bigr)}}}q(d\chi)\,dr\\ & = \int_s^t \zeta(r)\int f{{{\bigl((\chi_{t - r})_{t\in{{\ensuremath{\mathbbm{r}}}}}\bigr)}}}q(d\chi)\,dr .
\end{split}\ ] ] the last step follows from the pointwise convergence @xmath557 as @xmath449 for every @xmath558 together with the dominated convergence theorem and from the markov property .
finally let @xmath541 be integrable .
approximate @xmath541 with simple functions @xmath559 .
applying lemma [ l : first_moment_estimate ] , it is straight forward to show that equation with @xmath541 replaced by @xmath560 converges to equation as @xmath561 .
[ l : vanishing_immigration_weak_process ] assume [ a : a1_n ] and [ a : hutzenthaler2009ejp ] . fix @xmath562 .
suppose that @xmath563\to[0,\infty)$ ] are square integrable and that @xmath531 as @xmath22 in the @xmath564-norm .
let @xmath565 , @xmath321 , be independent solutions of satisfying @xmath566 for every @xmath48 .
then we have that @xmath567 where @xmath346 is a poisson point process on @xmath568\times u$ ] with intensity measure @xmath569 moreover let @xmath213 be a continuous function on @xmath570,{{\ensuremath{\mathbbm{r}}}}\bigr)}}}$ ] satisfying the lipschitz condition @xmath571,{{\ensuremath{\mathbbm{r}}}}\bigr)}}}\ ] ] for some @xmath572 , some @xmath130 and some @xmath573 .
in addition let @xmath216 be a continuous function satisfying @xmath217 for all @xmath164 .
then we have that @xmath574 } } } = { { \ensuremath{\mathbbm{e}}}}{{{\biggl[{{\ensuremath{\bar{f}}}}{{{\biggl(\bigl(\int { { \ensuremath{\bar{f}}}}{{{\bigl(\eta_{t - u}\bigr)}}}\pi(du , d\eta)\bigr)_{s\leq t\leq t}\biggr)}}}\biggr]}}}. \end{split}\ ] ] let @xmath575 be a countable dense subset of @xmath576 .
tightness of @xmath577 follows from tightness of @xmath578 this type of argument has been established in theorem 2.1 of roelly - coppoletta ( 1986 ) for the weak topology and @xmath579 . following the proof hereof
, one can show the analogous argument for the vague topology and @xmath580 .
fix @xmath581 and define @xmath582 for all @xmath583 and @xmath48 . note that @xmath212 is globally lipschitz continuous . for @xmath393 and fixed @xmath309 ,
global lipschitz continuity of @xmath212 implies that @xmath584 } } } \leq \frac{l_f}{k } \sup_{m\in{{\ensuremath{\mathbbm{n}}}}}m{{\ensuremath{\mathbbm{e}}}}^0{{{\biggl[y_{t , s}^{m,{\ensuremath{\hat{\zeta}}}}(1)\biggr]}}}\ ] ] for some constant @xmath585 and for all @xmath48 .
the right - hand side is finite according to lemma [ l : first_moment_estimate ] and converges to zero as @xmath383 .
this proves tightness of @xmath586 , @xmath48 , for every @xmath309 . for the second part of the aldous criterion , fix @xmath205 and let @xmath587 , @xmath48 , be stopping times which are uniformly bounded by @xmath588 .
in addition define @xmath589,n\in{{\ensuremath{\mathbbm{n}}}}.\ ] ] the functions @xmath159 , @xmath160 and @xmath176 are uniformly globally lipschitz continuous on the support of @xmath212 according to assumption [ a : a1_n ] .
therefore there exists a constant @xmath590 such that @xmath591 and @xmath592 for all @xmath164 and all @xmath48 . for fixed @xmath593 and @xmath594 $ ] , we use it s formula to obtain that @xmath595}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\bigl(\sum_{i=1}^n\int_{\tau_n}^{\tau_n+{{\ensuremath{\delta } } } } { { \ensuremath{{{\ensuremath{\mathcal g}}}}}}_{u}^n f{{{\bigl(y_{u , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\bigr)}}}\,du + \sum_{i=1}^n\int_{\tau_n}^{\tau_n+{{\ensuremath{\delta } } } } { { { \bigl(f^{'}{\ensuremath{{\displaystyle \cdot}}}\sigma_n\bigr)}}}{{{\bigl(y_{u , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\bigr)}}}db_u(i)\bigr)^2\biggr]}}}\\ & \leq 3c_f^2{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\bigl(\int_0^{{{\ensuremath{\delta}}}}{\ensuremath{\hat{\zeta}}}_n(\tau_n+u)du\bigr)^2\biggr ] } } } + 3c_f^2{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\bigl(\sum_{i=1}^n \int_0^{{\ensuremath{\delta}}}y_{\tau_n+u , s}^{n,{\ensuremath{\hat{\zeta}}}}(i ) du \bigr)^2\biggr]}}}\\ & \qquad\quad\qquad\qquad + 3\sum_{i=1}^n{{\ensuremath{\mathbbm{e}}}}{{{\bigl[\int_{0}^{{{\ensuremath{\delta } } } } { { { \bigl(f^{'}\sigma_n\bigr)}}}^2{{{\bigl(y_{\tau_n+u , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\bigr)}}}du\bigr]}}}\\ & \leq 3c_f^2{{\ensuremath{\delta}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\int_{\tau_n}^{\tau_n+{{\ensuremath{\delta } } } } \bigl({\ensuremath{\hat{\zeta}}}_n(u)\bigr)^2du\biggr ] } } } + 3c_f^2{{\ensuremath{\delta}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\int_0^{{\ensuremath{\delta}}}\bigl(\sum_{i=1}^n y_{\tau_n+u , s}^{n,{\ensuremath{\hat{\zeta}}}}(i ) \bigr)^2 du\biggr]}}}\\ & \qquad\quad\qquad\qquad + 3c_f^2\sum_{i=1}^n{{\ensuremath{\mathbbm{e}}}}{{{\bigl[\int_{0}^{{{\ensuremath{\delta } } } } \bigl(y_{\tau_n+u , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\bigr)^2 du\bigr]}}}\\ & \leq { { \ensuremath{\bar{{{\ensuremath{\delta}}}}}}}{\ensuremath{{\displaystyle \cdot}}}3c_f^2\sup_{m\in{{\ensuremath{\mathbbm{n}}}}}\int_s^{t+1}{{{\bigl({\ensuremath{\hat{\zeta}}}_m(u)\bigr)}}}^2 du + { { \ensuremath{\bar{{{\ensuremath{\delta}}}}}}}{\ensuremath{{\displaystyle \cdot}}}6c_f^2\sup_{m\in{{\ensuremath{\mathbbm{n } } } } } { { \ensuremath{\mathbbm{e}}}}^0{{{\biggl[\sup_{s\leq u\leq t+1}\bigl(\sum_{i=1}^m y_{u , s}^{m,{\ensuremath{\hat{\zeta}}}}(i)\bigr)^2\biggr ] } } } \end{split}\ ] ] for all @xmath596 and for all @xmath48 . in the last step we used that @xmath597 for every integrable function @xmath598 . the right - hand side of is finite by lemma [ l : y_2_supsum ] . letting @xmath599
, the left - hand side of converges to zero uniformly in @xmath48 and @xmath600 .
next we prove convergence of finite - dimensional distributions .
let @xmath130 , @xmath581 , @xmath601 and @xmath602 be arbitrary .
using independence we obtain that @xmath603 } } } = { { \ensuremath{\mathbbm{e}}}}{{{\bigl[e^{-\sum_{i=1}^n\sum_{j=1}^{n } { { \ensuremath{\lambda}}}_j f{{{\bigl(y_{t_j , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\bigr)}}}}\bigr]}}}\\ & = \prod_{i=1}^n{{{\bigl [ { { \ensuremath{\mathbbm{e}}}}e^{-\sum_{j=1}^{n } { { \ensuremath{\lambda}}}_j f{{{\bigl(y_{t_j , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\bigr)}}}}\bigr]}}}\\ & = \biggl(1-\frac{n { { \ensuremath{\mathbbm{e}}}}^0{{{\bigl [ 1-e^{-\sum_{j=1}^{n } { { \ensuremath{\lambda}}}_j f{{{\bigl(y_{t_j , s}^{n,{\ensuremath{\hat{\zeta}}}}(1)\bigr)}}}}\bigr]}}}}{n}\biggr)^n \end{split}\ ] ] for all @xmath48 . note that @xmath604 for all @xmath605 .
applying lemma [ l : vanishing_immigration_general ] to each summand of this telescope sum we get that @xmath606}}}\\ & = \exp{{{\biggl(-\int\int_s^\infty\zeta(r ) { { { \bigl(1-e^{-\sum_{j=1}^{n } { { \ensuremath{\lambda}}}_j f{{{\bigl(\eta_{t_j - r}\bigr)}}}}\bigr)}}}dr q(d\eta)\biggr)}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl [ \exp{{{\biggl(-\sum_{j=1}^{n}{{\ensuremath{\lambda}}}_j \int f{{{\bigl(\eta_{t_j - u}\bigr)}}}\pi(du , d\eta)\biggr)}}}\biggr]}}}. \end{split}\ ] ] this proves convergence of finite - dimensional distributions .
it remains to prove , which includes non - bounded test functions . by the previous step and by the skorokhod representation of weak convergence ( e.g. theorem ii.86.1 in @xcite )
there exists a version of @xmath607 which converges almost surely as @xmath22 . for every @xmath608 ,
let @xmath609 $ ] be a continuous function which satisfies @xmath610 for every @xmath611 $ ] and @xmath612 for every @xmath613 $ ]
. then @xmath614 .
thus @xmath615 converges almost surely as @xmath22 for every @xmath616 .
the second moment estimate of lemma [ l : y_2_supsum ] together with @xmath617 for all @xmath164 implies that the argument of @xmath213 in is almost surely uniformly bounded . consequently taking minimum with @xmath618
has no effect for sufficiently large @xmath616 and with @xmath619 converges almost surely , too .
uniform integrability of with @xmath619 follows from the @xmath402-estimate @xmath620}}}\\ & \leq 2{{\ensuremath{\bar{f}}}}{{{\bigl({{\ensuremath{\underline{0}}}}\bigr)}}}+ 2l_{{\ensuremath{\bar{f}}}}^2 l_{{\ensuremath{\bar{f}}}}^2{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\biggl(\sum_{j=1}^n \sum_{i=1}^n y_{t_j , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\biggr)^2\biggr]}}}\\ & \leq 2{{\ensuremath{\bar{f}}}}{{{\bigl({{\ensuremath{\underline{0}}}}\bigr)}}}+ 2l_{{\ensuremath{\bar{f}}}}^2 l_{{\ensuremath{\bar{f}}}}^2 n^2{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sup_{s\leq t\leq t } \big(\sum_{i=1}^n y_{t , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\big)^2\biggr ] } } } \end{split}\ ] ] and lemma [ l : y_2_supsum ] .
thus with @xmath619 converges also in expectation .
the lipschitz condition implies that @xmath621 } } } -{{\ensuremath{\mathbbm{e}}}}{{{\biggl[{{\ensuremath{\bar{f}}}}{{{\biggl(\bigl(\sum_{i=1}^n { { { \bigl({{\ensuremath{\bar{f}}}}{\ensuremath{{\displaystyle \cdot}}}h_k\bigr ) } } } { { { \bigl(y_{t , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\bigr)}}}\bigr)_{s\leq t\leq t}\biggr)}}}\biggr ] } } } \biggr|\\ & \leq\sum_{j=1}^n l_{{\ensuremath{\bar{f}}}}l_{{\ensuremath{\bar{f}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n y_{t_j , s}^{n,{\ensuremath{\hat{\zeta}}}}(i){{{\bigl(1-h_k{{{\bigl(y_{t_j , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\bigr)}}}\bigr)}}}\biggr]}}}\\ & \leq\sum_{j=1}^n l_{{\ensuremath{\bar{f}}}}l_{{\ensuremath{\bar{f}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n y_{t_j , s}^{n,{\ensuremath{\hat{\zeta}}}}(i ) { { { \bigl({{\ensuremath{\mathbbm{1}}}}_{\sup_{s\leq t\leq t}{{\ensuremath{|y_{t , s}^{n,{\ensuremath{\hat{\zeta}}}}|}}}\geq k } + { { \ensuremath{\mathbbm{1}}}}_{y_{t_j , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\leq\frac{1}{k } } \bigr ) } } } \biggr]}}}\\ & \leq\frac{n l_{{\ensuremath{\bar{f}}}}l_{{\ensuremath{\bar{f}}}}}{k}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sup_{s\leq t\leq t}{{\ensuremath{\bigl|y_{t , s}^{n,{\ensuremath{\hat{\zeta}}}}\bigr|}}}^2\biggr ] } } } + nl_{{\ensuremath{\bar{f}}}}l_{{\ensuremath{\bar{f}}}}\sup_{s\leq t\leq t}{{\ensuremath{\mathbbm{e}}}}{{{\bigl[\sum_{i=1}^n y_{t , s}^{n,{\ensuremath{\hat{\zeta}}}}(i)\wedge\frac{1}{k}\bigr ] } } } \end{split}\ ] ] for all @xmath48 . letting @xmath22 and then @xmath383 , the right - hand side converges to zero .
thus @xmath622 } } } } \\ & = { { \ensuremath{{\displaystyle \lim_{k { \rightarrow}\infty}}}}}{{\ensuremath{{\displaystyle \lim_{n { \rightarrow}\infty}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[{{\ensuremath{\bar{f}}}}{{{\biggl(\bigl(\sum_{i=1}^n { { { \bigl({{\ensuremath{\bar{f}}}}{\ensuremath{{\displaystyle \cdot}}}h_k\bigr ) } } } { { { \bigl(y_{t , s}(i)\bigr)}}}\bigr)_{s\leq t\leq t}\biggr)}}}\biggr]}}}\\ & = { { \ensuremath{{\displaystyle \lim_{k { \rightarrow}\infty}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[{{\ensuremath{\bar{f}}}}{{{\biggl(\bigl ( \int { { { \bigl({{\ensuremath{\bar{f}}}}{\ensuremath{{\displaystyle \cdot}}}h_k\bigr)}}}{{{\bigl(\eta_{t - u}\bigr)}}}\pi(du , d\eta)\bigr)_{s\leq t\leq t}\biggr)}}}\biggr]}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl[{{\ensuremath{\bar{f}}}}{{{\biggl(\bigl(\int { { \ensuremath{\bar{f}}}}{{{\bigl(\eta_{t - u}\bigr)}}}\pi(du , d\eta)\bigr)_{s\leq t\leq t}\biggr)}}}\biggr]}}}. \end{split}\ ] ] the last equality follows from the dominated convergence theorem together with assumption [ a : hutzenthaler2009ejp ] .
this proves .
recall the loop - free @xmath20-island process from .
the following lemma shows that the loop - free @xmath20-island process converges to the virgin island model .
[ l : convergence_of_the_loop_free_process ] assume [ a : a1_n ] , [ a : hutzenthaler2009ejp ] , [ a : initial ] and [ a : second_moments ] .
then we have that @xmath623 for every @xmath205 . fix @xmath205 .
let @xmath575 be a countable dense subset of @xmath576 .
tightness of the left - hand side of in @xmath48 follows from tightness of @xmath624 this type of argument has been established in theorem 2.1 of roelly - coppoletta ( 1986 ) for the weak topology and @xmath579 . following the proof hereof
, one can show the analogous argument for the vague topology and @xmath580 .
fix @xmath581 and define @xmath625 for all @xmath381 and @xmath48 . for fixed @xmath309 ,
global lipschitz continuity of @xmath212 implies that @xmath626 } } } \leq \frac{l_f}{k } \sup_{m\in{{\ensuremath{\mathbbm{n}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^m\sum_{k=0}^\infty z_{t}^{m , k}(i)\biggr]}}}\ ] ] for some constant @xmath585 and for all @xmath627 .
the right - hand side is finite according to lemma [ l : first_moment_estimate_x_z ] .
this proves tightness of for every fixed time point . for
the second part of the aldous criterion , let @xmath587 , @xmath48 , be stopping times which are uniformly bounded by @xmath588 .
in addition define @xmath628 and @xmath629 for all @xmath630 .
assumption [ a : a1_n ] implies that the functions @xmath159 , @xmath160 and @xmath176 are uniformly globally lipschitz continuous on the support of @xmath212 .
moreover @xmath631 is bounded by @xmath632 uniformly in @xmath48 .
therefore there exists a constant @xmath633 be such that @xmath634 for all @xmath635 and all @xmath48 and such that @xmath636 for all @xmath89 and @xmath48 . for fixed @xmath593 and @xmath594 $ ] , we use it s formula to obtain that @xmath637}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\bigl(\int_{\tau_n}^{\tau_n+{{\ensuremath{\delta } } } } { { \ensuremath{{{\ensuremath{\mathcal g}}}}}}^n h{{{\bigl(z_{u}^{n,\cdot}(\cdot)\bigr)}}}\,du + \sum_{i=1}^n\sum_{k=0}^\infty\int_{\tau_n}^{\tau_n+{{\ensuremath{\delta } } } } { { { \bigl(f^{'}{\ensuremath{{\displaystyle \cdot}}}\sigma_n\bigr)}}}{{{\bigl(z_{u}^{n , k}(i)\bigr)}}}db_u^k(i)\bigr)^2\biggr]}}}\\ & \leq 2c_h^2{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\bigl(\int_0^{{{\ensuremath{\delta } } } } 2{{\ensuremath{\theta}}}+\sum_{i=1}^n\sum_{k=0}^\infty z_{\tau_n+u}^{n , k}(i)\,du\bigr)^2\biggr ] } } } + 2\sum_{i=1}^n\sum_{k=0}^\infty{{\ensuremath{\mathbbm{e}}}}{{{\bigl[\int_{0}^{{{\ensuremath{\delta } } } } { { { \bigl(f^{'}\sigma_n\bigr)}}}^2{{{\bigl(z_{\tau_n+u}^{n , k}(i)\bigr)}}}du\bigr]}}}\\ & \leq { { \ensuremath{\bar{{{\ensuremath{\delta}}}}}}}{\ensuremath{{\displaystyle \cdot}}}4c_h^2\sup_{m\in{{\ensuremath{\mathbbm{n}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sup_{t\leq t+1 } \bigl(2{{\ensuremath{\theta}}}+\sum_{i=1}^m\sum_{k=0}^\infty z_{t}^{m , k}(i)\bigr)^2\biggr ] } } } \end{split}\ ] ] for all @xmath596 and for all @xmath48 .
the second step follows from @xmath638 for @xmath639 and from it s isometry . in the last step we used that @xmath597 for every integrable function @xmath598 .
the right - hand side of is finite according to lemma [ l : second_moment_estimate_x_z ] . letting @xmath599 , the left - hand side of converges to zero uniformly in @xmath48 and @xmath600 .
tightness of now follows from the aldous criterion , see aldous ( 1978 ) .
it remains to identify the limit of the left - hand side of by proving that @xmath640 for all @xmath641 , for all @xmath642 and for all @xmath643 .
lemma [ l : essentially_finitely_many_generations ] justifies to restrict the summation over @xmath310 to finitely many summands .
we prove by induction on @xmath644 using the poisson limit for independent one - dimensional diffusions .
define @xmath645 for every @xmath646,{{\ensuremath{\mathbbm{r}}}})$ ] .
note that @xmath209 satisfies the lipschitz condition for some constant @xmath527 .
let @xmath647 , @xmath68 , and @xmath648 , @xmath68 , be independent solutions of with @xmath649 , @xmath650 and @xmath651 for every @xmath48 and @xmath68 .
in addition let @xmath652 , @xmath68 , be independent solutions of with @xmath653 .
note that @xmath654 is a solution of with @xmath655 .
the first moment estimate of lemma [ l : first_moment_estimate ] implies that @xmath656 } } } - { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\biggl(-\sum_{i=1}^n f{{{\bigl({{{\bigl(y_{t,0}^{n,\zeta}(i)\bigr)}}}_{t\geq0}\bigr)}}}\biggr)}}}\biggr]}}}\bigr|}}}\\ & \leq l_f c_{t_n}\sum_{j=1}^n { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n{{\ensuremath{\bigl|x_0^n(\pi_i^n)-x_0(i)\bigr|}}}\biggr ] } } } \end{split}\ ] ] for all @xmath48 where @xmath657 is the finite constant of lemma [ l : first_moment_estimate ] . letting @xmath22
the right - hand side converges to zero according to assumption [ a : initial ] .
the process @xmath658 in turn is close to @xmath659 except for islands with a significant amount of mass at time zero . formalizing this we use lemma [ l : first_moment_estimate ] to obtain that @xmath660 } } } - { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\biggl(-\sum_{i=1}^n f{{{\bigl({{{\bigl({{\ensuremath{\tilde{y}}}}_{t,0}^{n,\zeta}(i)\bigr)}}}_{t\geq0}\bigr)}}}\biggr)}}}\biggr]}}}\bigr|}}}\\ & \leq { { \ensuremath{{\displaystyle \lim_{k { \rightarrow}\infty}}}}}l_f n c_{t_n}\sum_{i = k+1}^\infty { { \ensuremath{\mathbbm{e}}}}{{{\bigl[x_0(i)\bigr ] } } } + { { \ensuremath{{\displaystyle \lim_{k { \rightarrow}\infty}}}}}{{\ensuremath{{\displaystyle \lim_{n { \rightarrow}\infty}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[1-\exp{{{\biggl(-\sum_{i=1}^k f{{{\bigl({{{\bigl({{\ensuremath{\tilde{y}}}}_{t,0}^{n,\zeta}(i)\bigr)}}}_{t\geq0}\bigr)}}}\biggr)}}}\biggr]}}}. \end{split}\ ] ]
the first summand on the right - hand side is zero according to assumption [ a : initial ] .
note that @xmath661 and @xmath662 as @xmath22 by assumption [ a : a1_n ] .
thus @xmath663 converges in distribution to the zero function as @xmath22 for every fixed @xmath68 .
consequently the second summand on the right - hand side is zero .
moreover @xmath664 converges in distribution to @xmath665 as @xmath22 for every fixed @xmath68 .
these observations imply that @xmath666}}}\\ & = { { \ensuremath{{\displaystyle \lim_{k { \rightarrow}\infty}}}}}{{\ensuremath{{\displaystyle \lim_{n { \rightarrow}\infty}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\biggl(-\sum_{i=1}^k f{{{\bigl({{{\bigl(y_{t,0}^{n,\zeta}(i)\bigr)}}}_{t\geq0}\bigr)}}}\biggr)}}}\biggr ] } } } { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\biggl(-\sum_{i = k+1}^n f{{{\bigl({{{\bigl(y_{t,0}^{n,\zeta}(i)\bigr)}}}_{t\geq0}\bigr)}}}\biggr)}}}\biggr]}}}\\ & = { { \ensuremath{{\displaystyle \lim_{k { \rightarrow}\infty}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\biggl(-\sum_{i=1}^k f{{{\bigl({{{\bigl(y_t(i)\bigr)}}}_{t\geq0}\bigr)}}}\biggr)}}}\biggr ] } } } { { \ensuremath{{\displaystyle \lim_{n { \rightarrow}\infty}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\biggl(-\sum_{i=1}^n f{{{\bigl({{{\bigl({{\ensuremath{\tilde{y}}}}_{t,0}^{n,\zeta}(i)\bigr)}}}_{t\geq0}\bigr)}}}\biggr)}}}\biggr]}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\biggl(-\sum_{i=1}^\infty f{{{\bigl({{{\bigl(y_t(i)\bigr)}}}_{t\geq0}\bigr)}}}\biggr)}}}\biggr ] } } } { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\biggl(-\!\!\sum_{(s,\eta)\in\pi^{\emptyset } } f{{{\bigl((\eta_{t - s})_{t\geq0}\bigr ) } } } \biggr ) } } } \biggr]}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\bigl(-\!\!\sum_{(\iota , s,\eta)\in{{\ensuremath{\mathcal v}}}^{(0 ) } } f{{{\bigl((\eta_{t - s})_{t\geq0}\bigr ) } } } \bigr ) } } } \biggr]}}}. \end{split}\ ] ] the last but one step follows from lemma [ l : vanishing_immigration_weak_process ] with @xmath667 and @xmath668 .
this proves in the base case @xmath669 . for the induction step @xmath670 note that a version of @xmath671 conditioned on @xmath672 , @xmath673 , is given by the one - dimensional diffusion @xmath674 with vanishing immigration .
thus we may realize @xmath671 by choosing a suitable version of @xmath675 , @xmath676 , and by independently sampling a version of @xmath674 whose driving brownian motion is independent of @xmath677 .
tightness together with the induction hypothesis implies that @xmath678 for every @xmath679 .
thanks to the skorokhod representation of weak convergence ( e.g. theorem ii.86.1 in @xcite ) , we may assume that the convergence in holds almost surely . as a consequence
we obtain that @xmath680 holds almost surely . using arguments from the proof of one
can deduce from this that @xmath681 holds almost surely where the total mass of the @xmath132-th generation of the virgin island model is defined as @xmath682 for every @xmath683 . together with continuity of @xmath684
, implies that @xmath685 that this functional is not bounded is remedied by a truncation argument .
now the main step of the proof is lemma [ l : vanishing_immigration_weak_process ] with @xmath686 and @xmath687 for all @xmath46 .
lemma [ l : vanishing_immigration_weak_process ] implies that @xmath688}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\bigl[\exp{{{\bigl(-\sum_{i=1}^n f{{{\bigl({{{\bigl(y_{t,0}^{n,\zeta } \bigr)}}}_{t\leq t } \bigr ) } } } \bigr ) } } } \big|{{{\bigl(z_\cdot^{m , m}\bigr)}}}_{m\in{{\ensuremath{\mathbbm{n } } } } } \bigr]}}}\\ & { \longrightarrow}{{\ensuremath{\mathbbm{e}}}}{{{\bigl[\exp{{{\bigl(-\int f{{{\bigl((\eta_{t - s})_{t\leq t}\bigr)}}}\pi^{(m+1)}(dx , d\eta)\bigr)}}}\big| { { \ensuremath{\mathcal v}}}^{(m)}\bigr]}}}{\ensuremath{\qquad\text{as } n\to\infty}}\end{split}\ ] ] almost surely . here
@xmath689 conditioned on @xmath690 is a poisson point process on @xmath144 with intensity measure @xmath691 } } } = v_s^{(m)}\,ds\otimes q(d\eta)\\ & = \!\!\!\sum_{(\iota , r,\psi)\in{{\ensuremath{\mathcal v}}}^{(m)}}\!\!\!\psi_{s - r}ds\otimes q(d\eta ) = \!\!\!\sum_{(\iota , r,\psi)\in{{\ensuremath{\mathcal v}}}^{(m)}}\!\!\!{{\ensuremath{\mathbbm{e}}}}\pi^{(\iota , r,\psi)}(ds , d\eta ) .
\end{split}\ ] ] due to this decomposition of @xmath689 , we may realize @xmath689 conditioned on @xmath690 as the independent superposition of @xmath692 .
in other words , @xmath689 is equal in distribution to the @xmath693st - generation of the virgin island model
. therefore we get that @xmath694 } } } } \\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl\{{{\ensuremath{{\displaystyle \lim_{n { \rightarrow}\infty}}}}}\exp{{{\bigl(-\sum_{k=0}^{m}\sum_{i=1}^n f{{{\bigl(z_{\cdot}^{n , k}(i)\bigr ) } } } \bigr ) } } } { { \ensuremath{\mathbbm{e}}}}{{{\bigl[\exp{{{\bigl(-\sum_{i=1}^n f{{{\bigl(z_{\cdot}^{n , m+1}(i)\bigr ) } } } \bigr)}}}\big|{{{\bigl(z_\cdot^{m , m}\bigr)}}}_{m\in{{\ensuremath{\mathbbm{n } } } } } \bigr ] } } } \biggr\}}}}\\ & = { { \ensuremath{\mathbbm{e}}}}\biggl\{\exp{{{\biggl(-\sum_{k=0}^m\sum_{(\iota , s,\eta)\in{{\ensuremath{\mathcal v}}}^{(k ) } } \!\!\ ! f{{{\bigl((\eta_{t - s})_{t\leq t}\bigr ) } } } \biggr ) } } } { { \ensuremath{\mathbbm{e}}}}{{{\bigl[\exp{{{\bigl(-\!\!\!\sum_{(\iota , s,\eta)\in{{\ensuremath{\mathcal v}}}^{(m+1)}}\!\!\ ! f{{{\bigl((\eta_{t - s})_{t\leq t}\bigr)}}}\bigr)}}}\big| { { \ensuremath{\mathcal v}}}^{(m ) } \bigr]}}}\biggr\}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\exp{{{\bigl(-\sum_{k=0}^{m+1}\sum_{(\iota , s,\eta)\in{{\ensuremath{\mathcal v}}}^{(k ) } } f{{{\bigl((\eta_{t - s})_{t\leq t}\bigr ) } } } \bigr ) } } } \biggr ] } } } \end{split}\ ] ] which proves and completes the proof of lemma [ l : convergence_of_the_loop_free_process ] . in this subsection , we show that the @xmath169-process with migration levels and the loop - free @xmath169-process are identical in the limit @xmath22 .
our proof formalizes the following intuition .
the individuals of a certain migration level are concentrated on essentially finitely many islands . that these finitely many islands are populated by migrants of a different migration level has a probability of order @xmath695 . as a consequence ,
all individuals on one fixed island have the same migration level in the limit @xmath22 .
this intuition is subject of lemma [ l : one_generation_per_island ] .
first we show that a generation can not be dispersed uniformly over all islands . to obtain this interpretation from the following lemma ,
assume @xmath696 for all @xmath321 and some time @xmath46 .
then the cutting operation in has no effect for @xmath20 large enough .
however it is clear that the total mass of all individuals with migration level @xmath310 does not tend to zero as @xmath22 .
thus @xmath696 can not be true .
[ l : concentration ] assume [ a : a1_n ] , [ a : initial ] and [ a : second_moments ]
. then any solution of satisfies @xmath697 } } } { \xrightarrow{{{\ensuremath{\delta}}}{\rightarrow}0}}0\ ] ] for all @xmath309 .
the assertion is also true if @xmath313 is replaced by @xmath698 .
fix @xmath205 and @xmath699 $ ] . according to lemma [ l : essentially_finitely_many_generations ] , for each @xmath115 , there exists a @xmath700 such that @xmath701 thus it suffices to prove convergence of every summand in .
in addition , if we forget the migration levels in the @xmath169-process with migration levels , then we obtain the @xmath20-island model . more formally , lemma [ l : decomposition ] shows that @xmath702 defines an @xmath20-island model . recall @xmath703 from and fix @xmath393 . according to lemma [ l : tau_theta ] it suffices to prove with expectation being restricted to the event @xmath704 for every @xmath48 .
now lemma 3.3 in @xcite shows that , on the event @xmath705 , @xmath706 is stochastically bounded above by @xmath462 .
hence we get for all @xmath707 , @xmath308 and @xmath708 that @xmath709}}}}\\ & \leq \sum_{i=1}^n{{\ensuremath{\mathbbm{e}}}}{{{\biggl[{{\ensuremath{\mathbbm{1}}}}_{{{\ensuremath{\bar{x}}}}_0^n(i)\leq{{\ensuremath{\tilde{\delta } } } } } { { { \bigl(x_t^{n , k}(i)\wedge{{\ensuremath{\delta}}}\bigr)}}}{{\ensuremath{\mathbbm{1}}}}_{{{\ensuremath{\tilde{\tau}}}}_k^n > t}\biggr ] } } } + { { \ensuremath{\delta}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n{{\ensuremath{\mathbbm{1}}}}_{{{\ensuremath{\bar{x}}}}_0^n(i)>{{\ensuremath{\tilde{\delta}}}}}\biggr]}}}\\ & \leq \sum_{i=1}^n{{\ensuremath{\mathbbm{e}}}}^{{{\ensuremath{\bar{x}}}}_0^n(i)\wedge{{\ensuremath{\tilde{\delta } } } } } { { { \bigl(y_{t,0}^{n , k+n\mu_n(0)}\wedge{{\ensuremath{\delta}}}\bigr ) } } } + \frac{{{\ensuremath{\delta}}}}{{{\ensuremath{\tilde{\delta}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\bigl[\sum_{i=1}^n{{\ensuremath{\bar{x}}}}_0^n(i)\bigr]}}}\\ & \leq n{{\ensuremath{\mathbbm{e}}}}^{0 } { { { \bigl(y_{t,0}^{n , k+n\mu_n(0)}\wedge{{\ensuremath{\delta}}}\bigr ) } } } + c_t\sum_{i=1}^n{{\ensuremath{\mathbbm{e}}}}{{{\bigl[{{{\bigl({{\ensuremath{\bar{x}}}}_0^n(i)\wedge{{\ensuremath{\tilde{\delta}}}}\bigr)}}}\bigr ] } } } + \frac{{{\ensuremath{\delta}}}}{{{\ensuremath{\tilde{\delta}}}}}\sup_{m\in{{\ensuremath{\mathbbm{n}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\bigl[\sum_{i=1}^m{{\ensuremath{\bar{x}}}}_0^m(i)\bigr ] } } } \end{split}\ ] ] for some constant @xmath710 .
the last step follows from lemma [ l : first_moment_estimate ] .
next we let @xmath22 and @xmath449 in . applying lemma [ l : vanishing_immigration_weak_process ] with @xmath711 and using assumption [ a : initial ] , we see that the limit of as @xmath22 and as @xmath449 is bounded above by @xmath712}}}\ ] ] for every @xmath393 and @xmath708 .
the first summand in is zero by the dominated convergence theorem and lemma [ l : finite_excursion_area ] .
the last summand converges to zero as @xmath713 by the dominated convergence theorem and assumption [ a : initial ] .
this completes the proof of .
the proof of with @xmath313 replaced by @xmath698 is similar now we prove that all individuals on a fixed island have the same migration level in the limit @xmath22 .
more precisely , we show that @xmath313 and @xmath714 can not be big at the same time for any @xmath321 or any @xmath308 . [
l : one_generation_per_island ] assume [ a : a1_n ] , [ a : initial ] and [ a : second_moments ]
. then any solution of satisfies @xmath715 for every @xmath205 .
the assertion is also true if @xmath313 is replaced by @xmath698 .
we begin with the case @xmath716 .
fix @xmath393 and recall @xmath379 from . by the uniform local lipschitz continuity of @xmath159 ,
there exists a finite constant @xmath717 such that @xmath718 for all @xmath719 and all @xmath48 . applying it s formula ,
we see that @xmath720 where @xmath721 and @xmath722 are suitable martingales for each @xmath321 and @xmath723 .
now take expectations to obtain that @xmath724 for every @xmath48 and @xmath381 .
note that the right - hand side is finite .
using gronwall s inequality , @xmath725 and @xmath726 , we conclude that @xmath727 for every @xmath48 . letting @xmath22 proves in the case @xmath716 . for the case
@xmath728 , apply it s formula in @xmath729 and use similar estimates as above .
the case @xmath730 is analogous to the case @xmath728 . knowing that there is asymptotically at most one generation on every island , we are now in a position to prove that @xmath731 and @xmath732 are close to each other . [ l : x_close_to_z ] assume [ a : a1_n_linear ] , [ a : initial ] and [ a : second_moments ] . for each @xmath48 ,
let @xmath733 be a solution of and let @xmath734 be the unique solution of with @xmath735 replaced by @xmath736 .
then @xmath737 for every @xmath309 .
for the time being , assume that @xmath176 and @xmath159 are uniformly globally lipschitz continuous and bounded .
the general case will later be handled by a stopping argument .
define @xmath738 and @xmath739 for all @xmath740 .
we first prove with @xmath741 replaced by @xmath742 and @xmath743 , respectively , separately . as @xmath744 is not differentiable , we will apply it s formula to @xmath745 being defined as follows .
let @xmath746 satisfy @xmath747 note that @xmath748 as @xmath561 .
for every @xmath130 , there exists a continuous function @xmath749 with support in @xmath750 such that @xmath751 note that @xmath752 for all @xmath753 and @xmath130 . with this , we define the function @xmath754 for every @xmath130 .
these functions satisfy @xmath755 , @xmath756 , @xmath757 , @xmath758 and @xmath759 as @xmath760 for every @xmath740 and @xmath130 . denote the difference process by @xmath761 for all @xmath7 , @xmath46 , @xmath308 and @xmath48 .
the dependence on @xmath20 is suppressed for the sake of a more compact notation . by definition of @xmath745 , @xmath762 for all @xmath740 and @xmath130 .
thus @xmath763 for all @xmath393 and @xmath48 .
the first summand on the right - hand side converges to zero as @xmath383 uniformly in @xmath48 according to lemma [ l : essentially_finitely_many_generations ] .
the last summand on the right - hand side converges to zero as @xmath22 according to lemma [
l : concentration ] . consequently @xmath764 to estimate the right - hand side
, we apply it s formula to obtain that @xmath765 for every @xmath308 .
now we simplify the last summand on the right - hand side using the assumption of @xmath176 begin linear .
this linearity implies that @xmath766 for @xmath332 and @xmath333 .
in addition take expectations , estimate @xmath767 and use @xmath768 for @xmath740 .
thus we have that @xmath769 } } } } \\ & \leq { { \ensuremath{\mathbbm{e}}}}{{{\bigl[\sum_{j=1}^n{{\ensuremath{\bigl| \delta_t^{k-1}(j ) \bigr|}}}\bigr ] } } } + { { \ensuremath{\mathbbm{e}}}}{{{\bigl[\sum_{i=1}^n{{\ensuremath{\bigl|{{\ensuremath{\tilde{\mu}}}}_n{{{\bigl(x_t^{n , k}(i)\bigr ) } } } -{{\ensuremath{\tilde{\mu}}}}_n{{{\bigl(z_t^{n , k}(i)\bigr ) } } } \bigr| } } } \bigr]}}}\\ & \ + { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n{{\ensuremath{\bigl| \frac{x_t^{n , k}(i ) { \ensuremath{{\displaystyle \cdot}}}{{\ensuremath{\tilde{\mu}}}}_n{{{\bigl(\sum_{m=0}^\infty x_t^{n , m}(i)\bigr ) } } } } { \sum_{m=0}^\infty x_t^{n , m}(i ) } -{{\ensuremath{\tilde{\mu}}}}_n{{{\bigl(x_t^{n , k}(i)\bigr ) } } } \bigr| } } } \biggr]}}}\\ & \ + { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n \frac{\phi_{n^2}^{''}{{{\bigl(\delta_t^{k}(i)\bigr)}}}}{2 } \left(\sqrt { \sigma_n^2{{{\bigl(x_t^{n , k}(i)\bigr ) } } } } -\sqrt { \sigma_n^2{{{\bigl(z_t^{n , k}(i)\bigr ) } } } } \right)^2 \biggr ] } } } \end{split}\ ] ] for every @xmath46 , @xmath308 and @xmath48 . the last but one summand on the right - hand side
is estimated as follows .
let @xmath121 . in case of @xmath770 , apply the inequality @xmath771 with @xmath772 and @xmath773 .
in case of @xmath774 , use @xmath775 for all @xmath776 . for
the last summand on the right - hand side , use @xmath777 and @xmath778 for @xmath779 .
moreover recall @xmath780 for all @xmath558 .
hence , we get that @xmath781 } } } } \\ & \leq { { \ensuremath{\mathbbm{e}}}}{{{\bigl[\sum_{j=1}^n { { \ensuremath{\bigl|\delta_t^{k-1}(j)\bigr|}}}\bigr ] } } } + l{{\ensuremath{\mathbbm{e}}}}{{{\bigl[\sum_{i=1}^n{{\ensuremath{\bigl| x_t^{n , k}(i)-z_t^{n , k}(i ) \bigr| } } } \bigr]}}}\\ & \ + 2l{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n { { \ensuremath{\mathbbm{1}}}}_{x_t^{n , k}(i)\geq{{\ensuremath{\delta } } } } \sum_{m\neq k}x_t^{n , m}(i ) \biggr]}}}\\ & + 2 l{{\ensuremath{\mathbbm{e}}}}{{{\biggl [ \sum_{i=1}^{n}{{{\bigl(x_t^{n , k}(i)\wedge{{\ensuremath{\delta}}}\bigr ) } } } \biggr ] } } } + { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n \frac{l{{{\bigl(\delta_t^k(i)\bigr)}}}^{+}}{n^2{{{\bigl(\delta_t^k(i)\bigr)}}}^{+ } } \biggr ] } } } \end{split}\ ] ] for every @xmath46 , @xmath308 and @xmath48 .
summing over @xmath782 leads to @xmath783 } } } } \\ & \leq ( 1+l){{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{k=0}^\infty\sum_{i=1}^n{{\ensuremath{\bigl|\delta_t^{k}(i)\bigr|}}}\biggr ] } } } + \frac{2l}{{{\ensuremath{\delta}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{k=0}^\infty\sum_{i=1}^n x_t^{n , k}(i ) \sum_{m\neq k}x_t^{n , m}(i ) \biggr]}}}\\ & + 2 l{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{k=0}^\infty \sum_{i=1}^{n}{{{\bigl(x_t^{n , k}(i)\wedge{{\ensuremath{\delta}}}\bigr ) } } } \biggr ] } } } + ( k+1)\frac{l}{n } \end{split}\ ] ] for every @xmath46 and @xmath627 .
the last three summands on the right - hand side of converge to zero uniformly in @xmath699 $ ] if we first let @xmath22 and then @xmath449 .
this follows from lemma [ l : one_generation_per_island ] and lemma [ l : concentration ] . after inserting into
, we see that @xmath784 for all @xmath699 $ ] .
note that the right - hand side is finite by lemma [ l : essentially_finitely_many_generations ] .
similarly we obtain with @xmath785 replaced by @xmath786 .
finally apply gronwall s inequality to arrive at .
next we consider functions @xmath176 and @xmath159 which are not globally lipschitz - continuous .
for each @xmath787 choose function @xmath788 and @xmath789 which agree with @xmath160 and @xmath159 , respectively , on @xmath790 $ ] and which are uniformly globally lipschitz continuous and uniformly bounded .
existence of such functions follows from the uniform local lipschitz continuity of @xmath176 and @xmath159 .
recall the stopping time @xmath379 , @xmath627 , from .
the process @xmath791 agrees with the loop - free @xmath792-process on the event @xmath705 .
furthermore the process @xmath793 agrees with an @xmath792-process with migration levels on @xmath705 .
therefore @xmath794 for every @xmath787 by the preceding step .
lemma [ l : tau_theta ] handles the event @xmath704 . this completes the proof of lemma [ l : x_close_to_z ] .
first we prove theorem [ thm : convergence ] under the additional assumption [ a : second_moments ] .
this will be relaxed later .
we begin with convergence of finite - dimensional distributions .
recall @xmath524 from .
let @xmath795 satisfy the lipschitz condition with lipschitz constant @xmath527 and let @xmath209 be bounded by @xmath796 . furthermore let the function @xmath211 have compact support in @xmath83 , let @xmath212 be bounded by @xmath710 and let @xmath212 be globally lipschitz continuous with lipschitz constant @xmath797 .
recall the @xmath798-process with migration levels from .
we will exploit below that all individuals on one island have the same migration level in order to show that @xmath799 assuming we now prove convergence of finite - dimensional distributions .
we may replace the @xmath169-process with migration levels in by the loop - free process because of lemma [
l : x_close_to_z ] and the lipschitz continuity of @xmath209 and @xmath212 .
hence and lemma [ l : x_close_to_z ] imply that @xmath800 the last equality is the convergence of the loop - free @xmath169-process to the virgin island model and has been established in lemma [ l : convergence_of_the_loop_free_process ] .
next we prove . according to lemma [ l : decomposition ]
if we ignore the migration levels in the @xmath169-process with migration levels , then we obtain a version of the @xmath169-process , that is , @xmath801 for proving we observe that @xmath802 for every sequence @xmath803 and every @xmath121 .
thus we get that @xmath804 } } } + c_f\sum_{i=1}^n\frac{1}{{{\ensuremath{\delta}}}^2 } { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{k=0}^\infty x_{t}^{n , k}(i ) \sum_{l\neq k } x_{t}^{n , l}(i)\biggr]}}}\\ & = : c(n,{{\ensuremath{\delta}}},t ) \end{split}\ ] ] for all @xmath48 , @xmath46 and @xmath121 .
the second summand on the right - hand side converges to zero as @xmath22 according to lemma [ l : one_generation_per_island ] .
the first summand on the right - hand side converges to zero as @xmath449 uniformly in @xmath48 according to lemma [ l : concentration ] . using
we obtain that @xmath805}}}\\ & \qquad+\sum_{j=1}^n c(n,{{\ensuremath{\delta}}},t_j ) + \sum_{j=1}^n{{\ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n\sum_{k=0}^\infty{{\ensuremath{\mathbbm{1}}}}_{x_{t_j}^{n , k}(i)<{{\ensuremath{\delta } } } } \bigl| f{{{\bigl(x_{t_j}^{n , k}(i)\bigr)}}}\bigr|\biggr]}}}\\ & \leq \frac{l_f}{{{\ensuremath{\delta}}}}\sum_{j=1}^n { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n\sum_{k=0}^\infty x_{t_j}^{n , k}(i ) \sum_{m\neq k}x_{t_j}^{n , m}(i)\biggr]}}}\\ & \qquad+\sum_{j=1}^n c(n,{{\ensuremath{\delta}}},t_j ) + l_f\sum_{j=1}^n { { \ensuremath{\mathbbm{e}}}}{{{\biggl[\sum_{i=1}^n\sum_{k=0}^\infty x_{t_j}^{n , k}(i)\wedge{{\ensuremath{\delta}}}\biggr ] } } } \end{split}\ ] ] for all @xmath48 and @xmath121 . letting first @xmath22 and then @xmath449
, the right - hand side converges to zero according to lemmas [ l : one_generation_per_island ] and [ l : concentration ] and according to the preceding step .
inserting this into proves . the next step is to prove tightness
this is analogous to the tightness proof in lemma [ l : vanishing_immigration_weak_process ] .
use the lemmas [ l : tau_theta ] and [ l : second_moment_estimate_x ] instead of lemma [ l : y_2_supsum ] .
so we omit this step .
it remains to prove theorem [ thm : convergence ] in the case when assumption [ a : second_moments ] fails to hold .
fix @xmath205 .
let @xmath806 be a bounded continuous function on @xmath201 .
it follows from assumption [ a : initial ] that @xmath807 converges in @xmath564 and thus also in distribution to @xmath808 . by the skorokhod representation theorem
there exists a version of @xmath809 such that @xmath807 converges almost surely to @xmath808 as @xmath22 .
now the previous step implies that @xmath810 } } } { \xrightarrow{n { \rightarrow}\infty}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[h\biggl(\biggl(\sum_{(\iota , u,\eta)\in{{\ensuremath{\mathcal v}}}}{{\ensuremath{\delta}}}_{\eta_{t - u } } \biggr)_{t\leq t}\biggr)|x_0(\cdot)\biggr]}}}\ ] ] almost surely .
taking expectations and applying the dominated convergence theorem results in @xmath811 } } } { \xrightarrow{n { \rightarrow}\infty}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl[h\biggl(\biggl(\sum_{(\iota , u,\eta)\in{{\ensuremath{\mathcal v}}}}{{\ensuremath{\delta}}}_{\eta_{t - u}}\biggr)_{t\leq t}\biggr)\biggr]}}}\ ] ] almost surely .
this finishes the proof of theorem [ thm : convergence ] .
as in section [ sec : convergence_to_vim ] , we proceed in several steps .
first we define the loop - free process .
let @xmath812 be the solution of @xmath813 where we agree on @xmath814 for @xmath46 and @xmath7 .
we will refer to this process as the loop - free @xmath10-process .
the main two steps in our proof of theorem [ thm : comparison ] are as follows .
lemma [ l : z_dominates_x ] below shows that the total mass of the @xmath10-process is dominated by the total mass of the loop - free @xmath10-process .
lemma [ l : v_dominates_z ] then proves that the total mass of the loop - free @xmath10-process is dominated by the total mass of the virgin island model .
our proof of lemma [ l : v_dominates_z ] exploits the hierarchical structure of the loop - free process .
note that conditioned on migration level @xmath815 , the islands with migration level @xmath310 are independent one - dimensional diffusions .
we prepare this in subsection [ ssec : decomposition of a one - dimensional diffusion with immigration into subfamilies ] by studying the one - dimensional time - inhomogeneous diffusion @xmath816 where @xmath817 and @xmath44 .
the path @xmath818 will later represent the mass immigrating from lower migration levels .
the core of the comparison result is the following generator calculation which manifests the intuition that separating mass onto different islands increases the total mass .
if @xmath819 is constant , then a formal generator of @xmath820 is @xmath821 where @xmath822 , see e.g. section 5.3 in @xcite . recall @xmath823 from .
[ l : generator_estimate ] assume [ a : a1 ] .
suppose that @xmath824 satisfy @xmath825 .
assume @xmath11 to be subadditive , that is , @xmath826 for all @xmath422 with @xmath827 .
let @xmath297 .
if @xmath12 is superadditive , then @xmath828 if @xmath12 is subadditive , then @xmath829 if @xmath12 is additive , then @xmath830 for @xmath831 , the first derivative is nonnegative and the second derivative is nonpositive .
thus @xmath832 this is inequality .
the proof of inequality is analogous .
if @xmath12 is additive , then @xmath833 and no property of @xmath834 is needed in the above calculation .
note that the operator on the right - hand side of is a formal generator of the superposition @xmath835 of two independent solutions of .
this follows from theorem 4.10.1 in @xcite .
we will lift inequality between formal generators to an inequality between the associated semigroups . for this
we use the integration by parts formula . for its formulation ,
let @xmath836 and @xmath837 be two generators associated with the semigroups @xmath838 and @xmath839 , respectively .
then , for @xmath309 , we have that @xmath840 if @xmath841 for all @xmath444 , see p. 367
in liggett ( 1985 ) .
the idea of using for a comparison is borrowed from cox et al .
( 1996 ) . as the generator inequality holds for functions in @xmath842 , we need to show that the semigroup of @xmath843 preserves @xmath842 . this is subject of the following subsection .
we write @xmath844 for @xmath845 and @xmath130 .
the @xmath279-th unit row vector is denoted as @xmath846 for every @xmath68 .
recall @xmath286 from .
[ l : delete ] for every @xmath130 and @xmath847 , we have that @xmath848 for all @xmath849 , all @xmath850 , all @xmath851 and all @xmath852 .
the reverse inequality holds if @xmath286 is replaced by @xmath287 .
the proof is by induction on @xmath852 .
the base case @xmath853 is trivial .
now assume that holds for some @xmath854 .
fix @xmath855 , @xmath856 and @xmath851 .
applying the induction hypothesis at location @xmath857 to the index tuple @xmath858 , we obtain that @xmath859 the last step is again the induction hypothesis .
[ l : double_argument ] let @xmath130 , @xmath510 and @xmath860 .
then the two functions @xmath861 and @xmath862 are elements of @xmath863 .
this is also true if @xmath864 is replaced by @xmath282 and @xmath865 , respectively .
the functions @xmath866 and @xmath218 are non - decreasing and either bounded or nonnegative .
it is clear that @xmath218 is again @xmath275-convex for @xmath273 and that @xmath866 is @xmath275-convex for @xmath867 .
it remains to prove @xmath868-convexity of @xmath866 for @xmath869 .
applying lemma [ l : delete ] at location @xmath870 to the index tuple @xmath871 , we obtain for all @xmath856 that @xmath872 that is , @xmath866 is @xmath868-convex .
[ l : preservation ] assume [ a : a1 ] .
let @xmath130 and @xmath873 .
then the function @xmath874 is an element of @xmath875 for every @xmath876 .
if @xmath11 is concave , then this property still holds if @xmath877 is replaced by @xmath878 and if @xmath285 is replaced by @xmath286 , respectively .
fix @xmath876 and @xmath130 .
we only prove the case of @xmath11 being concave and @xmath879 as the remaining cases are similar . according to lemma [ l : double_argument ]
, it suffices to prove that @xmath880 is an element of @xmath881 .
let @xmath882 , @xmath164 , be solutions of with respect to the same brownian motion .
it is known that @xmath883 holds almost surely for all @xmath884 , see e.g. theorem v.43.1 in @xcite for the time - homogeneous case .
thus the function @xmath866 is again non - decreasing .
moreover @xmath866 inherits @xmath275-convexity from @xmath212 for every @xmath273 .
it remains to show that @xmath866 is @xmath885-convex for @xmath886 .
if @xmath887 , then @xmath885-convexity of @xmath212 at the point @xmath888 shows that @xmath889 for every @xmath856 and @xmath890 , that is , @xmath885-convexity of @xmath866 in the case @xmath887 .
one can establish convexity of @xmath891 as in lemma 6.1 of @xcite ( this lemma 6.1 shows concavity if @xmath212 is @xmath892-concave and smooth ) .
this step uses concavity of @xmath11 .
consequently , @xmath866 is @xmath892-convex .
this completes the proof of @xmath893 .
lemma [ l : preservation ] extends proposition 16 of cox et al .
( 1996 ) @xcite .
this proposition 16 is used in @xcite to establish a comparison result between diffusions with different diffusion functions , see theorem 1 in @xcite . using the above lemma [ l : preservation ]
, this comparison result can be extended to more general test functions .
feller s branching diffusion with immigration can be decomposed into independent families which originate either from an individual at time zero or from an immigrant , see e.g. theorem 1.3 in li and shiga ( 1995 ) .
a diffusion does in general not agree with its family decomposition if individuals interact with each other , e.g. if the branching rate depends on the population size .
if the drift function is subadditive and if the branching function is superadditive , however , then we get at least a comparison result .
in that situation , the diffusion is dominated by its family decomposition .
more precisely , the total mass increases in the order @xmath289 if we let all subfamilies evolve independently , see lemma [ l : family_decomposition ] below .
the following lemma is a first step in this direction .
[ l : semigroup_estimate ] assume [ a : a1 ] .
let @xmath297 and let @xmath894 be locally lebesgue integrable . if @xmath11 is concave and @xmath12 is superadditive , then @xmath895 if @xmath11 is concave and @xmath12 is subadditive , then inequality holds with @xmath284 replaced by @xmath283 .
if @xmath11 is subadditive and @xmath12 is additive , then inequality holds with @xmath284 replaced by @xmath268 .
let @xmath896 where @xmath897 .
we begin with the case of @xmath898 being simple functions . w.l.o.g .
we consider @xmath542 and @xmath899 where @xmath900 , @xmath544 and @xmath901 as we may let @xmath493 depend trivially on further time points .
we will prove by induction on @xmath130 that @xmath902 for the base case @xmath499 additionally assume @xmath903 .
approximate @xmath904 and @xmath11 with functions @xmath905 having the following properties .
all derivatives @xmath906,@xmath907 , @xmath852 , are bounded , @xmath908 is concave and @xmath909 is superadditive .
both functions vanish at zero . if @xmath76 , then @xmath910 .
moreover @xmath911 and @xmath912 as @xmath913 for all @xmath164 .
let @xmath914 be a solution of with @xmath12 and @xmath11 replaced by @xmath909 and @xmath908 , respectively , and let @xmath915 be an independent version hereof starting in @xmath89 .
then @xmath916 is twice continuously differentiable for every @xmath428 , see theorem 8.4.3 in gikhman and skorokhod ( 1969 ) . in addition
, lemma [ l : preservation ] proves @xmath917 for all @xmath918 $ ] .
consequently , we may apply lemma [ l : generator_estimate ] to @xmath919 for every @xmath920 $ ] and the integration by party formula yields that @xmath921 now as @xmath913 , @xmath922 converges weakly to @xmath923 for every @xmath164 , see lemma 19 in cox et al .
( 1996 ) for a sketch of the proof . therefore letting @xmath913 in
proves for @xmath499 if @xmath924 . the case of general
@xmath925 follows by approximating @xmath926 with smooth functions in @xmath842 . for the induction step @xmath927 ,
define @xmath928 note that the induction hypothesis implies that @xmath929 and that lemma [ l : preservation ] implies that @xmath930 .
therefore , using the markov property and the induction hypothesis , we get that @xmath931}}}\biggr]}}}\\ & = { { \ensuremath{\mathbbm{e}}}}f_{n+1}{{{\biggl({{{\bigl(y_{t , s}^{\zeta , x}+{{\ensuremath{\tilde{y}}}}_{t , s}^{{\ensuremath{\tilde{\zeta}}},y}\bigr)}}}_{t\geq s}\biggr ) } } } \end{split}\ ] ] for all @xmath422 satisfying @xmath827 .
the last step follows from the markov property and from independence of the two processes .
this proves . in case of general functions @xmath541 and @xmath932 , approximate @xmath541 and @xmath932 with simple functions @xmath933 and @xmath934 , @xmath935 , respectively .
the process @xmath936 converges in the sense of finite - dimensional distributions in @xmath564 , see lemma [ l : second_moment_estimate_y ] , and due to tightness also weakly to the process @xmath937 .
this completes the proof as the remaining cases are analogous .
[ l : small_initial_mass_weak ] assume [ a : a1 ] and [ a : hutzenthaler2009ejp ] .
then we have that @xmath938 for all @xmath776 and all @xmath562 where @xmath346 is a poisson point process on @xmath144 with intensity measure @xmath939 .
the proof is analogous to the proof of lemma [ l : vanishing_immigration_weak_process ] .
for convergence of finite - dimensional distributions use the convergence instead of lemma [ l : vanishing_immigration_weak_process ] .
tightness follows from an estimate as in together with boundedness ( see lemma 9.9 in @xcite ) of second moments .
finally we prove the main result of this subsection .
the following lemma shows that the total mass increases if we let all subfamilies evolve independently . in the special case of @xmath11 and @xmath12 being linear , inequality
is actually an equality according to the classical family decomposition of feller s branching diffusion with immigration .
[ l : family_decomposition ] assume [ a : a1 ] .
let @xmath940 be locally lebesgue integrable and let @xmath164 . if the drift function @xmath11 is concave and the diffusion function @xmath12 is superadditive , then @xmath941 for every @xmath44 where @xmath346 is a poisson point process on @xmath144 with intensity measure @xmath939 and where @xmath942 is an independent poisson point process on @xmath144 with intensity measure @xmath943 .
if @xmath11 is concave and @xmath12 is subadditive , then holds with @xmath284 replaced by @xmath283 . if @xmath11 is subadditive and @xmath12 is additive , then holds with @xmath284 replaced by @xmath268
the idea is to split the initial mass and the immigrating mass into smaller and smaller pieces .
fix @xmath44 .
let @xmath11 be concave and let @xmath12 be superadditive . according to lemma [ l : semigroup_estimate ] @xmath944 for every @xmath48 where all processes are independent of each other . letting @xmath22 in
, the right - hand side of converges to the right - hand side of , see lemma [ l : vanishing_immigration_weak_process ] and lemma [
l : small_initial_mass_weak ] .
the remaining cases are analogous .
[ l : z_dominates_x ] assume [ a : a1 ] .
if @xmath11 is concave and @xmath12 is superadditive , then @xmath945 if @xmath11 is concave and @xmath12 is subadditive , then inequality holds with @xmath284 replaced by @xmath283 . if @xmath11 is subadditive and @xmath12 is additive , then inequality holds with @xmath284 replaced by @xmath268 .
assume that @xmath11 is subadditive and that @xmath12 is superadditive .
we follow the proof of lemma [ l : semigroup_estimate ] and begin with a generator calculation similar to lemma [ l : generator_estimate ] .
let @xmath946 and @xmath947 denote the formal generators of @xmath10-process and of the loop - free @xmath10-process , respectively .
assume that @xmath948 depends only on finitely many coordinates .
associated with this test function is @xmath949 where @xmath950 .
note that @xmath951 .
the first partial derivatives @xmath952 , @xmath7 , are nonnegative and the second partial derivatives @xmath953 are nonpositive .
thus we see that @xmath954 } } } + \frac12\sum_{i\in g}f_{ii}{\ensuremath{{\displaystyle \cdot}}}\sigma^2{{{\bigl(\sum_{k=0}^\infty x_i^{(k)}\bigr)}}}\\ & \leq \sum_{k=0}^\infty{{{\biggl [ \sum_{i\in g}f_i{{{\bigl ( \sum_{j\in g}x_j^{(k)}m(j , i)-x_i^{(k ) } + \mu{{{\bigl(x_i^{(k)}\bigr ) } } } \bigr ) } } } + \frac12\sum_{i\in g}f_{ii}{\ensuremath{{\displaystyle \cdot}}}\sigma^2{{{\bigl ( x_i^{(k ) } \bigr ) } } } \biggr]}}}\\ & = \sum_{k=0}^\infty{{{\biggl [ \sum_{i\in g}f_i{{{\bigl ( \sum_{j\in g}x_j^{(k-1)}m(j , i)-x_i^{(k ) } + \mu{{{\bigl(x_i^{(k)}\bigr ) } } } \bigr ) } } } + \frac12\sum_{i\in g}f_{ii}{\ensuremath{{\displaystyle \cdot}}}\sigma^2{{{\bigl ( x_i^{(k ) } \bigr ) } } } \biggr]}}}\\ & = { { \ensuremath{{{\ensuremath{\mathcal g}}}}}}^z { { \ensuremath{\tilde{f}}}}{{{\bigl({{{\bigl(x_i^{(k)}\bigr)}}}_{i\in g , k\in{{\ensuremath{\mathbbm{n}}}}_0}\bigr ) } } } \end{split}\ ] ] for every @xmath950 .
now we wish to apply the integration by parts formula . in order to guarantee @xmath955 ,
approximate @xmath212 with smooth functions in @xmath842 , approximate @xmath11 and @xmath118 as in the proof of lemma [ l : semigroup_estimate ] and approximate @xmath5 with finite sets .
moreover in order to exploit the generator inequality in the integration by parts formula , we note that @xmath956 ( see lemma 6.1 in @xcite ) .
therefore the integration by parts formula together with inequality implies that @xmath957 for all @xmath950 .
in addition note that @xmath9 is stochastically non - decreasing in its initial configuration , see lemma 3.3 in @xcite .
as stochastic monotonicity is the only input into the proof of lemma [ l : preservation ] , the assertion of lemma [ l : preservation ] holds also for @xmath9 .
thus , for all @xmath958 , we have that @xmath959 using this , the assertion follows as in the proof of lemma [ l : semigroup_estimate ] by induction on the number of arguments of @xmath960 .
we show in lemma [ l : v_dominates_z ] below that the total mass of the loop - free process is dominated by the total mass of the virgin island model . in the proof of this lemma
, we use that the poisson point processes appearing in the definition of the virgin island model preserve convexity in a suitable way .
this is subject of the following lemma .
[ l : preservation_ppp ] for every vector @xmath961 , @xmath962 , let @xmath963 be a poisson point process on a polish space @xmath433 with intensity measure @xmath964 where @xmath965 are fixed measures on @xmath433 .
if @xmath966 , @xmath130 , then the function @xmath967 is an element of @xmath968 for every measurable test function @xmath969 .
analogous results hold if @xmath282 is replaced by @xmath864 and @xmath865 , respectively .
the function @xmath866 is still non - decreasing in the first @xmath132 variables and @xmath275-convex for all @xmath273 .
furthermore @xmath866 is non - decreasing in the last @xmath448 variables as @xmath963 is stochastically non - decreasing in @xmath970 .
fix @xmath971 and @xmath856 .
let the poisson point processes @xmath963 , @xmath972 and @xmath973 be independent .
fix @xmath974 , @xmath975 and a measurable test function @xmath969 .
note that @xmath976 has the same distribution as @xmath977 .
therefore , using @xmath892-convexity of @xmath212 in the point @xmath978 , we obtain that @xmath979 for the last step , note that @xmath980 has the same distribution as @xmath981 and that @xmath982 has the same distribution as @xmath983 .
this proves @xmath984-convexity of @xmath866 for all @xmath971 .
similar arguments prove @xmath985-convexity for @xmath869 and @xmath986 .
recall the total mass process @xmath987 of the @xmath132-th generation of the virgin island model from for every @xmath683 .
[ l : v_dominates_z ] assume [ a : a1 ] and [ a : migration ] .
if @xmath11 is concave and @xmath12 is superadditive , then @xmath988 for every @xmath989 .
if @xmath11 is concave and @xmath12 is subadditive , then inequality holds with @xmath284 replaced by @xmath283 .
if @xmath11 is subadditive and @xmath12 is additive , then inequality holds with @xmath284 replaced by @xmath268 .
we prove by induction on @xmath989 .
the base case @xmath990 follows from @xmath991 .
we apply lemma [ l : family_decomposition ] for the induction step @xmath992 .
fix @xmath993 where @xmath994 and @xmath995 .
let @xmath996 , @xmath997 , be independent poisson point processes on @xmath144 , independent of @xmath998 and with @xmath999 where @xmath1000 has intensity measure @xmath1001 note that conditioned on @xmath1002 , the law of @xmath1003 is equal to the law of @xmath1004 ( defined in ) where @xmath1004 , @xmath7 , are independent of each other . thus conditioning on @xmath998 and applying lemma [ l : family_decomposition ] with @xmath1005 and @xmath340 , we obtain that @xmath1006 } } } } \\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl\{{{\ensuremath{\mathbbm{e}}}}{{{\biggl[f_{n+1}{{{\bigl({{{\bigl({{\ensuremath{\bigl|z^{(k)}\bigr|}}}\bigr)}}}_{k=0,\ldots , k_0 } , \sum_{i\in g } y_{\cdot,0}^{\zeta_i^{(k_0)},0}(i)\bigr ) } } } \bigm|{{{\bigl(z^{(k)}\bigr)}}}_{k=0\ldots k_0 } \biggr ] } } } \biggr\}}}}\\ & \leq { { \ensuremath{\mathbbm{e}}}}{{{\biggl [ f_{n+1}{{{\bigl({{{\bigl({{\ensuremath{\bigl|z^{(k)}\bigr|}}}\bigr)}}}_{k=0,\ldots , k_0 } , \sum_{i\in g } \int_0^\infty \int\eta_{\cdot - u}\pi^{i,\zeta_i^{(k_0)}}(du , d\eta ) \bigr ) } } } \biggr]}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl [ f_{n+1}{{{\bigl({{{\bigl({{\ensuremath{\bigl|z^{(k)}\bigr|}}}\bigr)}}}_{k=0,\ldots , k_0 } , \int_0^\infty \int\eta_{\cdot - u}\pi^{\sum_{i\in g } \zeta_i^{(k_0)}}(du , d\eta ) \bigr ) } } } \biggr]}}}\\ & \leq { { \ensuremath{\mathbbm{e}}}}{{{\biggl [ f_{n+1}{{{\bigl({{{\bigl({{\ensuremath{\bigl|z^{(k)}\bigr|}}}\bigr)}}}_{k=0,\ldots , k_0 } , \int_0^\infty \int\eta_{\cdot - u}\pi^{{{\ensuremath{|z^{(k_0)}|}}}}(du , d\eta ) \bigr ) } } } \biggr]}}}. \end{split}\ ] ] the last inequality follows from @xmath1007 , where we used @xmath1008 from assumption [ a : migration ] , and where we used that @xmath1009 is non - decreasing .
next we would like to apply the induction hypothesis .
however the right - hand side of depends on @xmath1010 through a continuum of time points and not only through finitely many time points . to remedy this , we approximate the poisson point process on the right - hand side of by approximating @xmath1010 with simple functions . for each @xmath1011 , choose a discretization @xmath1012 of @xmath1013 $ ] of maximal width @xmath1014 such that @xmath1015 , @xmath1016 and @xmath1017 .
define @xmath1018 if @xmath1019 and @xmath1020 otherwise . for a path @xmath1021 ,
define @xmath1022 note that @xmath1023 for every @xmath46 as @xmath1024 .
thus the intensity measure @xmath1025 converges to @xmath1026 as @xmath1024 .
this convergence of the intensity measures implies weak convergence of the poisson point process @xmath1027 to the poisson point process @xmath1028 . due to lemma
[ l : preservation_ppp ] , the function @xmath1029 defined through @xmath1030 is an element of @xmath1031 .
now we apply the induction hypothesis and obtain that @xmath1032}}}\\ & = { { \ensuremath{{\displaystyle \lim_{m { \rightarrow}\infty}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl [ f{{{\bigl({{{\bigl({{\ensuremath{\bigl|z^{(k)}\bigr|}}}\bigr)}}}_{k=0,\ldots , k_0 } , \int_0^\infty \int\eta_{\cdot - u}\pi^{d_m{{\ensuremath{|z^{(k_0)}|}}}}(du , d\eta ) \bigr ) } } } \biggr]}}}\\ & \leq{{\ensuremath{{\displaystyle \lim_{m { \rightarrow}\infty}}}}}{{\ensuremath{\mathbbm{e}}}}{{{\biggl [ f{{{\bigl({{{\bigl({v^{(k)}}\bigr)}}}_{k=0,\ldots , k_0 } , \int_0^\infty \int\eta_{\cdot - u}\pi^{d_m{v^{(k_0)}}}(du , d\eta ) \bigr ) } } } \biggr]}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl [ f{{{\bigl({{{\bigl({v^{(k)}}\bigr)}}}_{k=0,\ldots , k_0 } , \int_0^\infty \int\eta_{\cdot - u}\pi^{{v^{(k_0)}}}(du , d\eta ) \bigr ) } } } \biggr]}}}\\ & = { { \ensuremath{\mathbbm{e}}}}{{{\biggl [ f{{{\bigl({{{\bigl({v^{(k)}}\bigr)}}}_{k=0,\ldots , k_0 } , v^{(k_0 + 1 ) } \bigr ) } } } \biggr]}}}. \end{split}\ ] ] putting and together completes the induction step .
we prove the case of @xmath11 being concave and @xmath12 being superadditive .
the remaining two cases are analogous . according to lemma
[ l : z_dominates_x ] we have that @xmath1033 for every finite subset @xmath1034 . letting @xmath1035
, we see that the total mass of the @xmath10-process is dominated by the total mass of the loop - free @xmath10-process .
now we get from lemma [ l : v_dominates_z ] that @xmath1036 for every @xmath700 . letting @xmath1037
, we obtain that the total mass of the loop - free @xmath10-process is dominated by the total mass of the virgin island model .
therefore , the total mass of the @xmath10-process is dominated by the total mass of the virgin island model .
i thank anton wakolbinger for inspiring discussions and valuable remarks .
mark kac . ,
volume 1957 of _ with special lectures by g. e. uhlenbeck , a. r. hibbs , and b. van der pol .
lectures in applied mathematics .
proceedings of the summer seminar , boulder , colo . _ interscience publishers , london - new york , 1959 .
henry p. mckean , jr .
propagation of chaos for a class of non - linear parabolic equations . in _
stochastic differential equations ( lecture series in differential equations , session 7 , catholic univ . , 1967 )
_ , pages 4157 .
air force office sci .
arlington , va . , 1967 .
etienne pardoux and anton wakolbinger . from exploration paths to mass excursions
variations on a theme of ray and knight . in
_ surveys in stochastic processes , proceedings of the 33rd spa conference in berlin , 2009 , j. blath , p. imkeller , s. roelly ( eds . ) , ems 2010 http://www.cmi.univ-mrs.fr/ pardoux / survey070410.pdf _ , 2010 . | we consider systems of interacting diffusions with local population regulation .
our main result shows that the total mass process of such a system is bounded above by the total mass process of a tree of excursions with appropriate drift and diffusion coefficients . as a corollary
, this entails a sufficient , explicit condition for extinction of the total mass as time tends to infinity . on the way to our comparison result
, we establish that systems of interacting diffusions with uniform migration between finitely many islands converge to a tree of excursions as the number of islands tends to infinity . in the special case of logistic branching
, this leads to a duality between the tree of excursions and the solution of a mckean - vlasov equation . |
inelastic scattering processes carry dynamical degrees of freedom which have a large influence on the electron dynamics .
signatures in the conductance of molecular electronics devices @xcite indicate coupling between electronic and vibrational degrees of freedom , whereas spin inelastic scattering effects have been utilized in experimental studies of the magnetic properties of magnetic atoms and clusters.@xcite a huge effort has been spent on studies of the influence of spin and vibrational degrees of freedom on the transport through quantum dots ( qds).@xcite studies of various aspects of inelastic scattering effects are of fundamental importance . recent developments towards incorporating superconducting electronics into nanoelectromechanical devices open possibilities of cooling @xcite and cooper pair shuttling.@xcite in this paper we focus on the dynamical aspects of a qd embedded in a josephson junction , which to our knowledge not has been studied previously . in this paper , we consider the influence of the mechanical motion of a molecular qd , embedded in a josephson junction , on the supercurrent flowing across the junction .
the mechanical motion of the qd couples to the electronic degrees of freedom of the tunneling electrons , which dramatically influence the dynamics of the electronic occupation in the qd .
we can clearly distinguish between six different regimes , with respect to the electron level @xmath0 and charging energy @xmath1 in the qd , see fig .
[ fig - pd ] , in which the dynamical aspects of the transport properties are different . in regimes
ii | vi , the dynamics of the qd properties generate new time - scales to the transport which are intimately associated with the energies of the single electron transitions in the qd . in regime
i , on the other hand , the internal properties of the qd are static , even for finite bias voltages . the motion and josephson current are , thus , set solely by the applied bias voltage and phase difference between the electrodes .
the paper is organized as follows . in sec .
[ sec - model ] we introduce the model of the josephson junction in which the molecular qd is embedded , and we derive the basic expressions for the josephson current , the two - electron tunneling process , and the occupation numbers in the qd . the results are analyzed in sec . [ sec - dynamics ] and we conclude the paper in sec .
[ sec - discussion ] . and charging energy @xmath1 . here , @xmath2 and @xmath3 , where @xmath4 , with @xmath5 , whereas @xmath6 is the superconducting gap in the electrodes.,scaledwidth=45.0% ]
we start by considering a molecular qd embedded in a josephson junction , where the qd is exposed to mechanical oscillations which are modeled with a hook s law constant @xmath7 , and we assume that there is a bias voltage applied across the junction .
we illustrate the system schematically in fig .
[ fig - system ] .
the hamiltonian for the set - up is expressed by @xmath8 where @xmath9 , @xmath10 , are usual @xmath11-wave bcs hamiltonians , whereas @xmath12 defines the qd , with single - electron levels @xmath4 which are spin split by the effective field @xmath13 , and with charging energy @xmath1 . finally , @xmath14 accounts for the single - electron tunneling between the lead @xmath15 and the qd with rate @xmath16
. the local vibrational mode of the island is in the linear coupling regime given by @xmath17 where @xmath18 describes the coupling between the tunneling electrons and the vibrational mode corresponding the to left ( right ) tunnel junction .
the quantity @xmath19 is the displacement operator for the oscillator .
the tunneling matrix element @xmath16 is exponential in the displacement @xmath19 , thus , the assumed linear coupling is a good approximation for small @xmath19 .
this allows evaluation of @xmath18 in terms of the tunneling matrix elements and their distance dependence .
we assume here a very general equilibrium geometry with no particular symmetry being required .
the equilibrium position ( @xmath20 ) for the qd within the junction may be placed anywhere in between the leads . and mass @xmath21 .
the device is biased with the voltage @xmath22.,scaledwidth=45.0% ] we derive the basic formula for the supercurrent through the system using an analogous procedure as in refs . .
the total current flowing between the qd and lead @xmath15 can be fundamentally expressed as @xmath23 , from which we obtain @xmath24 here , we have defined the current operator @xmath25 using that @xmath26 and analogously for the @xmath27-operators , where @xmath28 and @xmath29 is the chemical potential and total number of electrons in lead @xmath15 , whereas we put the chemical potential for the qd to zero , i.e. @xmath30 . the total current comprise one component of single - particle , or electron , current which is contained in the first term of the above expression , while the second term describes the supercurrent . in the present paper , we are interested in the supercurrent only and , therefore , we discard the first contribution to the total current in the following discussion .
the average @xmath31 is decoupled into the averages e.g. @xmath32 and @xmath33 of which the former is further handled via e.g. bogoliubov - valatin transformation @xmath34 , such that ( @xmath35 ) @xmath36 , \\
\calf^{\dagger,<}_{\bfk\sigma\bar\sigma}(t , t')\equiv & i\av{\csdagger{\bfk\bar\sigma}(t')\cdagger{\bfk}(t ) } \nonumber\\= & i\sigma u_\bfk^*v_\bfk [ f(-e_\bfk)e^{ie_\bfk\tau}-f(e_\bfk)e^{-ie_\bfk\tau } ] .\end{aligned}\ ] ] here , @xmath37 defines the quasi - particle energies , whereas @xmath38/e_\bfk)/2}$ ] and @xmath39/e_\bfk)/2}$ ] , which satisfy @xmath40 and @xmath41 , with the macroscopic phase @xmath42 .
we shall proceed at low temperatures , such that we can approximate @xmath43 and @xmath44 .
the average @xmath33 is handled by transforming the qd hamiltonian into diagonal form using the expansion @xmath45 for the operators @xmath46 , @xmath2 , @xmath47.@xcite , such that @xmath48 , @xmath3 .
then , since @xmath49 , we define the anomalous averages for the qd according to @xmath50 where @xmath51 is the average rate for the two - electron transition @xmath52 at the energy @xmath53 . using the prescribed procedure
, we find that the supercurrent between the superconducting lead @xmath15 and the qd at low temperatures can be written @xmath54 e^{-i\omega\tau}dt'\frac{d\omega}{2\pi } , \label{eq - ismain}\end{aligned}\ ] ] where @xmath55 , and where we have assumed the local approximation @xmath56 , which is justified since the vibrational motion is much slower than the electronic tunneling processes .
the amplitudes @xmath57 and @xmath58 of the josephson current in absence and presence of the coupling to the vibrational mode are given by @xmath59 , \\
\gamma_\chi(\mu_\chi)=&2e\sum_{k\sigma\in\chi}\frac{|\delta_\chi||t_\chi^{(0)}|^2}{2e_k } [ \call_{\bfk-}^2(\omega+\delta_{\bar\sigma0 } ) \nonumber\\ & -\call_{\bfk+}^2(\omega+\delta_{2\sigma})],\end{aligned}\ ] ] respectively , where @xmath60)$ ] , @xmath61 . the quadratic dependence of the displacement @xmath19 in the current , eq .
( [ eq - ismain ] ) , is justified since the @xmath62 for small @xmath19 .
the quadratic component , thus , merely provides a minor modification to the linear displacement .
the dynamics of the average @xmath63 is calculated through the equation of motion @xmath64 where the correlation function is treated by using perturbation theory , which to first order in @xmath16 gives @xmath65(t)}= \nonumber\\&\hspace{-3.5cm}= i\int_{-\infty}^tt_\chi ( [ g_{0\sigma}^>(t , t ' ) + g_{\bar\sigma2}^>(t , t')]{\cal f}^<_{k\sigma\bar\sigma}(t',t ) \nonumber\\&\hspace{-3 cm } -[g_{0\sigma}^<(t , t')+g_{\bar\sigma2}^<(t , t')]{\cal f}^>_{k\sigma\bar\sigma}(t',t ) ) e^{-i\mu_\chi t'}dt'.\end{aligned}\ ] ] the lesser and greater gfs for the qd , are here given by @xmath66 where @xmath67 , denotes the occupation number for the qd state @xmath68 . whereas the anomalous gfs for the leads are expressed as @xmath69 , \\ { \cal f}_{k\sigma\bar\sigma}^{<}(t , t')\equiv & i\av{\cs{k\bar\sigma}(t')\cc{k}(t ) } \nonumber\\= & i\sigma u_kv_k^*[f(e_\bfk)e^{-ie_k\tau}-f(-e_\bfk)e^{ie_\bfk\tau}].\end{aligned}\ ] ] the equation for @xmath70 is , thus , given by @xmath71\}e^{-i\omega(t - t')}\frac{d\omega}{2\pi}dt'e^{-i(\omega_\chi t+\phi_\chi ) } , \label{eq - n02}\end{aligned}\ ] ] where the amplitudes @xmath72 and @xmath73 are @xmath74 \\ v_\chi(\omega , t)=&\sum_{k\sigma\in\chi}\frac{|\delta_\chi||t_\chi^{(0)}|^2}{2e_k } [ n_0\call_{\bfk-}^2(\delta_{\sigma0}-\omega ) \nonumber\\ & + n_{\bar\sigma}\call_{\bfk-}^2(\delta_{2\bar\sigma}-\omega ) + n_\sigma\call_{\bfk+}^2(\delta_{\sigma0}-\omega ) \nonumber\\ & + n_2\call_{\bfk+}^2(\delta_{2\bar\sigma}-\omega ) ] .\end{aligned}\ ] ] those amplitudes describe the cotunneling processes in which two electrons are either added or removed from the qd in absence and presence of the coupling to the vibrational mode , respectively .
it is clear that the supercurrent is to a great extent determined by the time - evolution of the electron occupation @xmath75 in the qd .
we obtain those from the density matrix @xmath76 requiring @xmath77 . within the employed level of approximation ,
@xmath78 is determined from the master equation @xmath79 , where @xmath80 , whereas @xmath81 and where @xmath82 here , @xmath83 is the density of electron states in lead @xmath15 , whereas @xmath84 .
it can be seen in tabs . [ tab - pd1 ] and [ tab - pd2 ] , and fig .
[ fig - pd ] that there are several regimes in the @xmath85-space , in which the time - dependence of the quantum dot occupation numbers is very different . here
, we analyze each of the regimes in order to elucidate the characteristic time - scales involved in the electro - mechanical dynamics of the qd .
.the three regimes in which at least one transition energy lies within the superconducting gap . [ cols="<,^,^,^ " , ] we note several things in this formula for the josephson current .
the first term would arise when the qd is rigid in space since there is no coupling to the vibrational motion .
all other terms , which are proportional to @xmath86 , arise due to the movability of the qd in the josephson junction .
there are two terms directly depending on the position @xmath19 of the qd , as expected from previous studies.@xcite more interesting , however , is the dependence on the velocity @xmath87 and acceleration @xmath88 of the qd , displayed in the fourth , fifth , and sixth terms .
these dependences , hence , open possibilities for novel measurements and applications of shuttling qds . following the procedure introduced in refs .
, we derive the effective hamiltonian @xmath89 by requiring that @xmath90 , where @xmath91 is given in eq .
( [ eq - jschi ] ) .
we briefly discuss the salient features of the result , while a detailed derivation will be presented elsewhere . within the given level of approximation ,
the hamiltonian is linear in the coordinates @xmath92 .
hence , the classical equation of motion from the hamiltonian @xmath93 is expected to be simple , in the sense that all features derived from the superconducting current act as a driving force @xmath94 .
the equation of the qd motion will , therefore , be that of a driven oscillator , i.e. @xmath95 , where the damping factor @xmath96 contains the external damping due to mechanical friction .
we notice _ en passant _ that a quadratic approximation in @xmath86 provides contributions that add a josephson stiffness and damping .
the linear spatial motion of the qd is , however , given by @xmath97h(\omega_\chi,\phi_\chi ) \nonumber\\ & + \sum_{\chi'\neq\chi}\biggl ( \calq_{\chi\chi'}(\omega_{\chi ' } ) h(\omega_{\chi'},\phi_{\chi ' } ) -\frac{1}{2}\calq_{\chi\chi'}(\omega_{\chi'}-\omega_\chi ) \nonumber\\&\times h(\omega_{\chi'}-\omega_\chi,\phi_{\chi'}-\phi_\chi ) -\frac{1}{2}\calq_{\chi\chi'}(\omega_{\chi'}+\omega_\chi ) \nonumber\\&\times h(\omega_{\chi'}+\omega_\chi,\phi_{\chi'}+\phi_\chi ) \biggr)\biggr\ } , \label{eq - q}\end{aligned}\ ] ] where @xmath98u_{\chi ' } ' + \tilde{\alpha}_{\chi'}\omega[2u_{\chi'}''-v_{\chi ' } ' ] , \\ h(\omega,\phi)=&\frac{1-(\omega/\omega_0)^2}{[1-(\omega/\omega_0)^2]^2+[\gamma_n\omega / k_c]^2 } \nonumber\\&\times \biggl [ \cos(\omega t+\phi ) + \frac{\gamma_n\omega}{k_c } \frac{\sin(\omega t+\phi)}{1-(\omega/\omega_0)^2 } \biggr],\end{aligned}\ ] ] whereas @xmath99 and @xmath100 are dimensionless parameters , whereas @xmath101 is the josephson energy .
the first term in eq .
( [ eq - q ] ) describes the unperturbed damped motion of the qd around its equilibrium position @xmath102 , where @xmath103 and @xmath104 are to be determined from the initial conditions , whereas @xmath105 ^ 2}$ ] ( @xmath106 ) is the eigenfrequency of the damped ( undamped ) mechanical oscillations .
the only time - scales introduced in the josephson current are the ones associated with the eigenfrequency @xmath107 and the josephson frequencies @xmath108 , which is contrasted by the properties of the other regimes , as we shall see below .
this extended analysis of the expected josephson current in term of the qd motion is possible only since the occupation numbers are constants of motion .
hence , for the remaining regimes we will be content with establishing the characteristic time - scales . within regime
ii , the parameters @xmath109 , for all transitions @xmath110 , and also @xmath111 since @xmath112 , whereas @xmath113 since @xmath114 .
hence , the number @xmath115 is a constant of motion , while the remaining system of equations has the solution @xmath116 for some initial time @xmath117 at which @xmath118 . under the local approximation of the tunneling rate , @xmath119 acquires the time - evolution @xmath120 [ 1-\alpha_\chi\dot{q}(t - t_0 ) ] + \alpha_\chi^2\dot{q}^2(t - t_0)^2/3 \ } ( t - t_0 ) } .\end{aligned}\ ] ] the exponent periodically changes sign with the qd velocity @xmath87 , and position @xmath19 , which provides an oscillatory behavior of the occupation numbers @xmath119 as well as for @xmath121 .
physically , this means that the electron density in the qd periodically grows and wanes as the qd moves , which opens for possibilities to load and unload electron density on the qd at different leads , or , single electron shuttling between the superconducting leads via the qd .
the properties of the qd in regime iii are analogously obtained by interchanging the roles of @xmath119 and @xmath115 .
the time - scale @xmath122 for loading ( unloading ) electron density on the qd is related to the transition energy @xmath123 ( @xmath124 ) , and it can be noted that @xmath125 as @xmath126 ( @xmath127 ) , while it is determined solely by the density of electron states , @xmath128 , in the leads for transition energies far below ( above ) the superconducting gap .
the dynamics of the occupation numbers introduce an additional time - scale to the ac josephson current which is associated with the energies of the qd transitions , and which is different from the one introduced through the josephson frequency . in regime iv , all transition energies @xmath129 , which implies that @xmath109 , whereas @xmath130 .
we can , then , integrate the occupation of the empty state according to eq .
( [ eq - n0i ] ) , whereas the occupations for the one- and two - electron states are given by @xmath131 while both the empty and one - electron states , again , depend on the periodic motion and velocity of the qd , the two - electron state also acquires an oscillatory behavior since it depends on the integrated time - evolution of the occupation of the other states .
it is noticeable , however , that the time - evolution of the electron occupation in the qd here is related to the ( four ) time - scales associated with the transition energies @xmath132 and @xmath133 .
in particular , the one - electron occupations @xmath121 strongly depend on the rate of the transitions @xmath134 and @xmath135 . the properties of regime vi are obtained by noticing that all @xmath136 and all @xmath137 such that the roles of @xmath119 and @xmath115 become interchanged . in regime
v , finally , @xmath114 and @xmath138 , which leads to that @xmath139 , @xmath140 , @xmath111 , and @xmath141 .
@xmath119 and @xmath115 are , thus , determined by eq .
( [ eq - n0i ] ) , requiring the replacements @xmath142 and @xmath143 in the expression for @xmath115 .
it is understood that @xmath119 decreases ( increases ) while @xmath115 increases ( decreases ) , which is expected from conservation of probability .
the one - electron occupations @xmath121 depend on the integrated time - evolution of @xmath119 and @xmath115 , that is , @xmath144 dt ' .\end{aligned}\ ] ] the occupation of the one - electron states can , thus , be viewed as resulting from the imbalance between the occupation in the empty and two - electron states
we have studied the dynamics of a single level molecular qd embedded in a josephson junction , in which the mechanical motion of the qd is couples to the supercurrent . in the static limit , i.e.
when the qd is rigid , the supercurrent between the superconducting electrodes is mediated via two - electron transitions in the qd , a behavior which remains also when the qd moves between the electrodes .
the rate of the two - electron transitions naturally depends on the occupation of the qd states and the energies of the transitions between the states .
the rate does , in addition , explicitly depend on the motion of the qd .
our main focus has , in the present paper , been devoted to extract the time - scales that are involved in the dynamics of the qd occupation numbers .
it turns out that we can clearly distinguish between six separate regimes in the phase space of the single electron energy level @xmath0 and the qd charging energy @xmath1 , see the phase diagram in fig .
[ fig - pd ] .
the boundaries between the various regimes are set by pairing potential ( gap function ) of the electrodes , and the dynamics of the qd occupation depends on whether the single - electron transitions in the qd lie within or outside the gap . in all regimes but one , there are more than two time - scales involved in the dynamics of the qd occupation .
those additional time - scales are typically defined by the energies of the single - electron transitions relative to the superconducting gap ( and the chemical potential ) . for transition energies very far below ( above ) the gap , the time - scale for occupying , or deoccupying ,
the corresponding states are set by the density of electron states in the electrodes , while the time - scale tends to zero as the transition energies approach the edge of the gap from below ( above ) .
as the transition energies lie within the gap function , the occupation numbers of the qd are constants of the motion which implies that the dynamics of the qd occupation depends solely on i ) the two basic time - scales set by the bias voltage ( josephson frequency ) and the eigenfrequency @xmath107 of the oscillator , and ii ) the phase difference between the electrodes .
an experimental set - up to test our predictions would be possible by having gap functions @xmath145 mev , which is pertinent to mgb@xmath146,@xcite in order to obtain a sufficiently small mechanical damping . by also acquiring a vibrational frequency of @xmath147
ghz,@xcite one should be able to tune the josephson frequency such that @xmath148 .
a coupling strength of @xcite @xmath149 should be sufficient for an efficient read - out .
finally , the charging energy @xmath1 of the qd may be of the order of 1 | 10 mev , while the level spacing should preferably be larger than @xmath1 in order to control the electron occupation . for this requirement on the charging energy
, one would have access to the regimes i | iv and vi , whereas the regime v would become accessible by requiring a charging energy @xmath150 .
j.f . acknowledges support from the swedish research council and royal swedish academy of sciences .
a.v.b . and j.x.z
acknowledge that this work was supported by us doe , ldrd and bes , and was carried out under the auspices of the nnsa of the us doe at lanl under contract no .
de - ac52 - 06na25396 .
20 v. j. langlais , r. r. schlittler , h. tang , a. gourdon , c. joachim , and j. k. gimzewski , phys .
rev . lett .
* 83 * , 2809 ( 1999 ) .
h. park , j. park , a. k. l. lim , e. h. anderson , a. p. alivisatos , and p. l. mceuen , nature ( london ) * 407 * , 57 ( 2000 ) .
r. h. m. smit , y. noat , c. untiedt , n. d. lang , m. van hemert , and j. m. van ruitenbeek , nature ( london ) * 419 * , 906 ( 2002 ) .
j. park , nature ( london ) * 417 * , 722 ( 2002 )
. j. reichert , r. ochs , d. beckmann , h. b. weber , m. mayor , and h. v. lohneysen , phys .
* 88 * , 176804 ( 2002 ) .
n. b. zhitenev , h. meng , and z. bao , phys . rev
. lett . * 88 * , 226801 ( 2002 ) .
p. gambardella , s. rusponi , m. veronese , s. s. dhesi , c. grazioli , a. dallmeyer , i. cabria , r. zeller , p. h. dederichs , k. kern , c. carbone , and h brune , science , * 300 * , 1130 ( 2003 ) .
a. j. heinrich , j. a. gupta , c. p. lutz , and d. m. eigler , science * 306 * , 466 ( 2004 ) . c. f. hirjibehedin , c. p. lutz , and a. j. heinrich , science * 312 * , 1021 ( 2006 ) . c. f. hirjibehedin , c. -y .
lin , a. f. otte , m. ternes , c. p. lutz , b. a. jones , and a. j. heinrich , science * 317 * , 1199 ( 2007 ) .
j. fransson , nano lett . * 9 * , 2414 ( 2009 ) .
t. balashov , t. schuh , a. f. takcs , a. ernst , s. ostanin , j. henk , i. mertig , p. bruno , t. miyamachi , s. suga , and w. wulfhekel , phys .
* 102 * , 257203 ( 2009 ) . a. v. rozhkov and d. p. arovas , phys .
lett . * 82 * , 2788 ( 1999 ) .
k. flensberg , phys . rev .
b * 68 * , 205323 ( 2003 ) ; j .- x .
zhu , z. nussinov , a. shnirman , and a. v. balatsky , phys .
lett . * 92 * , 107001 ( 2004 ) .
f. siano and r. egger , phys .
* 93 * , 047002 ( 2004 ) .
j. fransson and j. -x .
zhu , phys .
b * 78 * , 113307 ( 2008 ) ; m. galperin , a. nitzan , and m. a. ratner , phys .
b * 78 * , 125320 ( 2008 ) .
a. naik , o. buu , m. d. lahaye , a. d. armour , a. a. clerk , m. p. blencowe , and k. c. schwab , nature ( london ) * 443 * , 193 ( 2006 ) .
j. p. pekola , f. giazotto , and o. -p .
saira , phys .
98 * , 037201 ( 2007 ) ; j. p. pekola and f. w. j. hekking , phys . rev .
_ ibid_. * 98 * , 210604 ( 2007 ) .
b. c. stipe , h. j. mamin , t. d. stowe , t. w. kenny , and d. rugar , phys .
* 86 * , 2874 ( 2001 ) .
x. m. h. huang , c. a. zorman , m. mehregany , and , m. l. roukes , nature * 421 * , 496 ( 2003 ) .
e. k. irish , j. gea - banacloche , i. martin , and k. c. schwab , phys .
b , * 72 * , 195410 ( 2005 ) . | we investigate dynamical transport aspects of a combined nanomechanical - superconducting device in which cooper pair tunneling interfere with the mechanical motion of a vibrating molecular quantum dot embedded in a josephson junction .
six different regimes for the tunneling dynamics are identified with respect to the electron level and the charging energy in the quantum dot . in five of those regimes
new time - scales are introduced which are associated with the energies of the single electron transitions within the quantum dot , while there is one regime where the internal properties of the quantum dot are static . |
deformation of the nuclear shape from spherical symmetric one has long been one of the most important issues in nuclear structure physics . among them , searches for evidences of the triaxial ( @xmath2 or @xmath0 ) one have been pursued long time , for example , the even - odd energy staggering in the low - spin part of the @xmath0 bands @xcite , the signature dependence of the energy spectra and the @xmath3 transition rates in medium - spin odd - odd and odd-@xmath4 nuclei @xcite , properties of the @xmath5 isomers @xcite , and so on .
but their results have not been conclusive ; making a clear distinction between the static and the dynamic ( vibrational ) ones has not been successful up to now .
theoretically , appearance of the wobbling motion , which is well - known in classical mechanics of asymmetric tops @xcite and whose quantum analog was discussed in terms of a rotor model about thirty years ago @xcite , is a decisive evidence of static triaxial deformations .
subsequently its microscopic descriptions were developed by several authors @xcite .
since the small - amplitude wobbling mode carries the same quantum numbers , parity @xmath6 and signature @xmath7 , as the odd - spin members of the @xmath0 band , ref.@xcite anticipated that it would appear as a high - spin continuation of the @xmath0 band , but it has not been resolved that in what nuclei , at what spins , and with what @xmath0 wobbling modes would be observed .
shimizu and matsuyanagi @xcite and onishi @xcite performed extensive numerical calculations for normally - deformed er isotopes with relatively small @xmath8 .
matsuzaki @xcite , shimizu and matsuzaki @xcite , and horibata and onishi @xcite also studied @xmath9os with relatively large negative-@xmath0 but their correspondence to experimental information has not been very clear .
these studies indicate the necessity of high - spin states in stably and strongly @xmath0-deformed nuclei .
bengtsson studied high - spin states around @xmath10hf @xcite and found systematic existence of the tsd ( triaxial super- or strongly deformed ) states with @xmath11 and @xmath12 .
this confirmed the discussion on the shell gap at @xmath13 in ref.@xcite , the work in which the yrast tsd band in @xmath14lu was reported ; and in 2000 an excited tsd band was observed in this nucleus and from the strengths of the interband @xmath15 transition rates this was unambiguously identified with the wobbling motion @xcite .
this data was analyzed by using a particle - rotor model @xcite and the @xmath15 transition rates were reproduced well . subsequently tsd bands were found in some lu and hf isotopes and wobbling excitations were observed also in @xmath16lu @xcite
. a close look at these data , however , tells us that the sign of their @xmath0-deformation seems to contradict to an irrotational motion and that the unexpected behavior of the wobbling frequency has not been explained yet .
thus in the preceding rapid communication @xcite we presented an answer to these problems . in the present paper , after summarizing the discussion there we extend numerical analyses to elucidate it .
an emphasis is put on the behavior of the calculated dynamic moments of inertia .
we start from a one - body hamiltonian in the rotating frame , @xmath17 in eq.([hsp ] ) , @xmath18 and 2 stand for neutron and proton , respectively , and chemical potentials @xmath19 are determined so as to give correct average particle numbers @xmath20 . the oscillator frequencies in eq.([hnil ] )
are expressed by the quadrupole deformation parameters @xmath21 and @xmath0 in the usual way .
they are treated as parameters as well as pairing gaps @xmath22 .
the orbital angular momentum @xmath23 in eq.([hnil ] ) is defined in the singly - stretched coordinates @xmath24 , with @xmath25 1 3 denoting @xmath26
@xmath27 , and the corresponding momenta . by diagonalizing @xmath28 at each @xmath29
, we obtain quasiparticle ( qp ) orbitals and the nuclear yrast ( 0qp ) state .
since @xmath28 conserves parity @xmath30 and signature @xmath31 , nuclear states can be labeled by them .
nuclear states with qp excitations are obtained by exchanging the qp energy and wave functions such as @xmath32 where @xmath33 denotes the signature partner of @xmath34 .
we perform the random phase approximation ( rpa ) to the residual pairing plus doubly - stretched quadrupole - quadrupole ( @xmath35 ) interaction between qps . since we are interested in the wobbling motion that has a definite quantum number , @xmath36 , only two components out of five of the @xmath35 interaction are relevant .
they are given by @xmath37 where the doubly - stretched quadrupole operators are defined by @xmath38 and those with good signature are @xmath39 the residual pairing interaction does not contribute because @xmath40 is an operator with @xmath41 .
the equation of motion , @xmath42_\mathrm{rpa } = \hbar\omega_n x_n^\dagger , \ ] ] for the eigenmode @xmath43 leads to a pair of coupled equations for the transition amplitudes @xmath44\right\rangle .\ ] ] then , by assuming @xmath45 , this can be cast @xcite into the form @xmath46 \notag \\ = 0 , \end{gathered}\ ] ] which is independent of @xmath47s .
this expression proves that the spurious ( nambu goldstone ) mode given by the first factor and all nomal modes given by the second are decoupled from each other . here
@xmath48 as usual and the detailed expressions of @xmath49 are given in refs.@xcite . among normal modes ,
one obtains @xmath50 by putting @xmath51 .
note that this gives a real excitation only when the right - hand side is positive and it is non - trivial whether a collective solution appears or not .
evidently this coincides with the form derived by bohr and mottelson in a rotor model @xcite and known in classical mechanics @xcite , aside from the crucial feature that the moments of inertia are @xmath29-dependent in the present case .
one drawback in our formulation is that our @xmath52 tends to be larger than corresponding experimental values because of the spurious velocity dependence of the nilsson potential as discussed in refs.@xcite .
a remedy for this was discussed there but that for @xmath53 has not been devised yet . therefore we assume for the present a similar discussion holds for the latter and accordingly the ratio @xmath54 which actually determines @xmath55 is more reliable than their absolute magnitudes .
interband electric quadrupole transitions between the @xmath56-th excited band and the yrast are given as @xmath57 in terms of @xmath58 they will be abbreviated to @xmath59 later for simplicity . in - band ones are given as @xmath60 in terms of @xmath61 and assumed to be common to all bands .
they will be abbreviated to @xmath62 .
here we adopted a high - spin approximation @xcite .
the transition quadrupole moment @xmath63 is extracted from @xmath62 by the usual rotor - model prescription . to compare collectivities of these two types of @xmath15 transitions
, we introduce a pair of deformation parameters @xmath64 then it is evident that the in - band one is expressed as @xmath65 as for the interband ones , by expanding @xmath66 by @xmath67s and @xmath68s , where @xmath56 runs both normal modes and the nambu
goldstone mode @xmath69 , we obtain from @xmath70=0 $ ] a kind of sum rule @xmath71 consecutively introducing the ratios of the dynamic to static deformations , @xmath72 the sum rule above reads @xmath73 the dynamic amplitudes @xmath74 describe shape fluctuations associated with the vibrational motion in the uniformly rotating frame .
transformation to the body - fixed ( principal - axis ) frame @xcite turns the shape fluctuation into the fluctuation of the angular momentum vector , i.e. , the wobbling motion .
this transformation relates the ratios , @xmath75 and @xmath76 , to the moments of inertia @xcite : @xmath77 where @xmath78 is a real amplitude that relates the dynamic amplitude @xmath74 and the moment of inertia , @xmath79 is the sign of @xmath80 ( so @xmath81 for wobbling - like rpa solutions ) , and @xmath82 thus , the interband @xmath83 is rewritten as , @xmath84 ^ 2 , \label{be2wob}\end{gathered}\ ] ] which coincides with the formula given by the rotor model @xcite , except for the appearance of the amplitude @xmath78 and sign @xmath79 . substituting the ratios , @xmath75 and @xmath76 , into eq.([sumrl ] ) , one finds that the amplitudes should satisfy @xmath85 this form of sum rule clearly indicates that the amplitude @xmath78 is a microscopic correction factor quantifying the collectivity of the wobbling motion , for which @xmath86 means the full - collectivity and reproduces the results of the macroscopic rotor model in both the energy and the interband @xmath83 values .
since the first experimental confirmation of the wobbling excitation in @xmath14lu @xcite , @xmath87 has been widely accepted as the shape of the tsd states in this region .
this is predominantly because the calculated energy minimum for @xmath87 is deeper than that for @xmath88 @xcite according to the shape driving effect of the aligned @xmath1 quasiparticle .
the recent precise measurements of @xmath63 @xcite also support this . on the other hand ,
the sign of @xmath0-deformation leads different consequences on moments of inertia , which are directly connected to the excitation energy of the wobbling mode through the wobbling frequency formula @xcite , c.f .
eq.([disp ] ) .
since the rpa is a microscopic formalism , no distinction between the collective rotation and the single - particle degrees of freedom has been made .
therefore , the moments of inertia calculated in our rpa formalism in sect.[sectwob ] are those for rotational motions of the whole system . in contrast , the macroscopic irrotational - like moments of inertia are often used in the particle - rotor calculations , where @xmath89 for @xmath87 and they lead to an imaginary wobbling frequency @xmath55 .
it is , however , noted that the moments of inertia of the particle - rotor model are those of the rotor and no effect of the single - particle alignments is included , so that they do not necessarily correspond to those calculated in our rpa formalism . in the preceding paper @xcite
we have performed microscopic rpa calculations without dividing the system artificially into the rotor and particles .
that work proved that for the calculated moment of inertia , @xmath48 , the contribution from the aligned qp(s ) , @xmath90 with @xmath91 being the aligned angular momentum , is superimposed on an irrotational - like moment of inertia ( @xmath92 ) of the core " .
consequently the total @xmath52 is larger than @xmath93 , which makes wobbling excitation in @xmath94 nuclei possible .
the second consequence of the formulation adopted in ref.@xcite is that the three moments of inertia are automatically @xmath29-dependent even when the mean - field parameters are fixed constant .
this is essential in order to explain the observed @xmath29-dependence of @xmath55 _ decreasing _ as @xmath29 increases .
otherwise @xmath55 is proportional to @xmath29 .
another important feature of the data is that the interband @xmath83 values between the wobbling and the yrast tsd bands are surprisingly large .
our rpa wave function gave extremely collective @xmath59 that gathered @xmath95 0.6 0.8 in the sum rule ( eq.([sum ] ) ) but the result accounted for only about one half of the measured one .
to elucidate these findings more , in the following we extend our numerical analyses putting a special emphasis on the @xmath0-dependence of the moments of inertia in subsect.[sectlow ] .
dependence on other parameters is also studied in detail .
features in common and different between even - even and odd-@xmath4 nuclei are also pointed out . in subsect.[secthigh ] , we discuss @xmath29-dependence . in subsect.[sectbe2 ] , characteristics of @xmath59 are discussed .
calculations are performed in five major shells ; @xmath96 3 7 for neutrons and @xmath96 2 6 for protons .
the strengths @xmath97 and @xmath98 in eq.([hnil ] ) are taken from ref.@xcite .
hafnium-168 is the first even - even nucleus in which tsd bands were observed @xcite . in this nucleus
three tsd bands were observed but interband @xmath0-rays connecting them have not been observed yet .
this means that the character of the excited bands has not been established , although we expect at least one of them is wobbling excitation .
an important feature of the data is that the average transition quadrupole moment was determined as @xmath101 eb .
this imposes a moderate constraint on the shape . referring to the weak parameter dependence discussed later
, we choose @xmath102 , @xmath103 , and @xmath104 0.3 mev , which reproduce the observed @xmath63 , as a typical mean - field parameter set .
first we study the dependence of various quantities on @xmath0 and other mean - field parameters at @xmath105 0.25 mev . around this frequency the @xmath106 alignment that is essential for making wobbling excitation in @xmath94 nuclei possible is completed and therefore the wobbling motion is expected to emerge above this frequency ( see fig.[fig7 ] shown later ) .
figure [ fig1 ] shows dependence on @xmath0 calculated with keeping @xmath102 and @xmath104 0.3 mev .
figure [ fig1](a ) graphs the calculated excitation energy in the rotating frame , @xmath107 . as @xmath0 comes close to 0
( symmetric about the @xmath27 axis ) and @xmath108 ( symmetric about the @xmath109 axis ) , @xmath55 approaches 0 , see eq.([disp ] ) .
we did not obtain any low - lying rpa solutions at around @xmath110 whereas a collective solution appears again for @xmath111 .
hf , calculated at @xmath105 0.25 mev with @xmath102 and @xmath104 0.3 mev .
[ fig1],title="fig:",width=264 ] hf , calculated at @xmath105 0.25 mev with @xmath102 and @xmath104 0.3 mev .
[ fig1],title="fig:",width=264 ] hf , calculated at @xmath105 0.25 mev with @xmath102 and @xmath104 0.3 mev .
[ fig1],title="fig:",width=264 ] hf , calculated at @xmath105 0.25 mev with @xmath102 and @xmath104 0.3 mev .
[ fig1],title="fig:",width=264 ] figure [ fig1](b ) shows the calculated moments of inertia .
their @xmath0-dependence resembles the irrotational , the so - called @xmath0-reversed , and the rigid - body moments of inertia , in @xmath112 , @xmath113 , and @xmath111 , respectively .
these model moments of inertia are given by @xmath114 @xmath115 and @xmath116 where @xmath25 1 3 denote the @xmath26
@xmath27 principal axes , @xmath117 the irrotational mass parameter , @xmath118 the rigid moment of inertia in the spherical limit , and @xmath119 is a deformation parameter like @xmath21 .
the @xmath0-reversed moment of inertia was introduced to describe positive-@xmath0 rotations in the particle - rotor model @xcite but its physical meaning has not been very clear ; in particular , it does not fulfill the quantum - mechanical requirement that the rotations about the symmetry axis should be forbidden .
we have clarified in the preceding paper @xcite that the contributions from aligned quasiparticles superimposed on irrotational - like moments of inertia ( @xmath120 ) can realize @xmath121 and this is the very reason why the wobbling excitation ( see eq.([disp ] ) ) appears in positive-@xmath0 nuclei .
we also discussed that multiple alignments could eventually lead to a rigid - body - like moment of inertia .
figure [ fig1](c ) indicates that , in the present calculation in which configuration is specified as the adiabatic quasiparticle vacuum at each @xmath29 , two @xmath1 protons align for @xmath94 as mentioned above while they have not fully aligned for @xmath122 at this @xmath29 . in other words , these figures cover the both regions in which the @xmath106 alignment is necessary ( @xmath123 ) and that is not necessary ( @xmath124 ) for obtaining wobbling excitations .
this aligned angular momentum determines the overall @xmath0-dependence of @xmath52 in fig.[fig1](b ) . as for the neutron part , corresponding to the disappearance of the solution at around @xmath110 , the expectation value of the neutron angular momentum , @xmath125 , drops around this region . to look at this more closely
, we investigate the nilsson single - particle diagram at @xmath126 .
figure [ fig2](a ) graphs neutron single - particle energies for @xmath127 with @xmath128 , while fig.[fig2](b ) for @xmath129 with @xmath102 .
the chemical potential that gives correct neutron number @xmath130 for @xmath94 at @xmath131 = 0.25 mev is also drawn in the latter .
this figure clearly shows that with this @xmath21 a shell gap exists for @xmath132 at @xmath130 . and by comparing this with fig.[fig1 ]
we see that the dropping of @xmath125 is a consequence of the deoccupation of the orbital that is [ 651 1/2 ] at @xmath128 ( hereafter simply referred to as the [ 651 1/2 ] orbital even at @xmath45 ) originating from the mixed ( @xmath133-@xmath134 ) spherical shell .
figure [ fig2](b ) also explains the reason why the wobbling excitation revives at around @xmath135 again ; the occupation of other oblate - favoring orbitals such as [ 503 7/2 ] makes it possible and leads to a rigid - body - like behavior of the moments of inertia .
figures [ fig2](c ) and ( d ) are corresponding ones for protons .
this indicates that the proton shell gap is robuster . , ( a ) for @xmath127 with @xmath128 , and ( b ) for @xmath129 with @xmath102 for neutrons .
( c ) and ( d ) are corresponding ones for protons .
solid and dashed curves represent even and odd parity orbitals , respectively .
asymptotic quantum numbers of some important orbitals are explicitly indicated .
chemical potentials that give particle numbers @xmath130 and @xmath136 for @xmath137 at @xmath131 = 0.25 mev are also indicated in ( b ) and ( d ) . [ fig2],title="fig:",width=264 ] , ( a ) for @xmath127 with @xmath128 , and ( b ) for @xmath129 with @xmath102 for neutrons .
( c ) and ( d ) are corresponding ones for protons .
solid and dashed curves represent even and odd parity orbitals , respectively .
asymptotic quantum numbers of some important orbitals are explicitly indicated .
chemical potentials that give particle numbers @xmath130 and @xmath136 for @xmath137 at @xmath131 = 0.25 mev are also indicated in ( b ) and ( d ) .
[ fig2],title="fig:",width=264 ] , ( a ) for @xmath127 with @xmath128 , and ( b ) for @xmath129 with @xmath102 for neutrons .
( c ) and ( d ) are corresponding ones for protons .
solid and dashed curves represent even and odd parity orbitals , respectively .
asymptotic quantum numbers of some important orbitals are explicitly indicated .
chemical potentials that give particle numbers @xmath130 and @xmath136 for @xmath137 at @xmath131 = 0.25 mev are also indicated in ( b ) and ( d ) .
[ fig2],title="fig:",width=264 ] , ( a ) for @xmath127 with @xmath128 , and ( b ) for @xmath129 with @xmath102 for neutrons .
( c ) and ( d ) are corresponding ones for protons .
solid and dashed curves represent even and odd parity orbitals , respectively .
asymptotic quantum numbers of some important orbitals are explicitly indicated .
chemical potentials that give particle numbers @xmath130 and @xmath136 for @xmath137 at @xmath131 = 0.25 mev are also indicated in ( b ) and ( d ) .
[ fig2],title="fig:",width=264 ] figure [ fig1](d ) graphs the quadrupole transition amplitudes @xmath138 ( @xmath139 ) associated with the wobbling mode .
( @xmath138 corresponds to @xmath140 in ref.@xcite . )
this shows that their relative sign changes with that of @xmath0 as discussed in refs.@xcite .
this feature can be understood as : @xmath141 is the @xmath0-vibrational region because the @xmath142 component is dominant ( see also @xmath143 and @xmath144 in fig.[fig1](b ) ) , and the mixing of the @xmath145 component due to triaxiality and rotation gives rise to the character of the wobbling motion .
this relative sign leads to a selection rule of the interband transition probabilities @xmath59 @xcite .
in the present case we obtain @xmath146 for @xmath147 , and typically their ratio to the in - band ones is @xmath148 .
figure [ fig3 ] shows dependence on @xmath21 calculated with keeping @xmath103 and @xmath104 0.3 mev .
the steep rises at around @xmath149 in figs.[fig3](a ) and ( b ) indicate the necessity of the @xmath106 ( the [ 660 1/2 ] orbital in fig.[fig2](c ) ) alignment for the appearance of the wobbling mode although the critical value of @xmath21 itself is frequency dependent . aside from this
, @xmath55 is almost constant in the calculated range .
the slight increase at around @xmath150 stems from the occupation of the @xmath151[651 1/2 ] orbital .
we have confirmed that in this case the @xmath152 alignment at around @xmath153 seen in fig.[fig3](b ) does not affect @xmath55 visibly since @xmath154 in this case is almost the same as @xmath155 although its reason is not clear .
figure [ fig3](c ) graphs @xmath63 .
this figure indicates that the chosen shape @xmath156 and @xmath103 reproduces the measured @xmath63 .
hf , calculated at @xmath105 0.25 mev with @xmath103 and @xmath104 0.3 mev .
[ fig3],title="fig:",width=264 ] hf , calculated at @xmath105 0.25 mev with @xmath103 and @xmath104 0.3 mev .
[ fig3],title="fig:",width=264 ] hf , calculated at @xmath105 0.25 mev with @xmath103 and @xmath104 0.3 mev .
[ fig3],title="fig:",width=264 ] figure [ fig4](a ) shows dependence on the pairing gaps . since we do not have detailed information about the gaps , we assume @xmath157 for simplicity .
this figure shows that the dependence on the gaps is weak unless they are too large . since the static pairing gap @xmath99 is expected to be small , say , @xmath158 0.6 mev , in the observed frequency range
, @xmath55 is not sensitive to the value of @xmath99 .
this is a striking contrast to the @xmath119 and @xmath0 vibrations ; it is well known that pairing gaps are indispensable for them . here
we note that the behavior of the @xmath55 correlates well with @xmath159 presented in fig.[fig4](b ) .
hf , calculated at @xmath105 0.25 mev with @xmath102 and @xmath103 .
@xmath157 is assumed for simplicity .
[ fig4],title="fig:",width=264 ] hf , calculated at @xmath105 0.25 mev with @xmath102 and @xmath103 .
@xmath157 is assumed for simplicity .
[ fig4],title="fig:",width=264 ] next we study @xmath160lu in a way similar to the preceding @xmath100hf case .
we choose @xmath103 and @xmath104 0.3 mev as representative mean - field parameters as above .
as for @xmath21 , however , we examined various possibilities because @xmath63 has not been measured in this nucleus .
since the sensitive @xmath21-dependence through the occupation of the @xmath151[651 1/2 ] orbital appears only at @xmath161 0.4 mev and therefore the band - head " properties do not depend on @xmath21 qualitatively , first we discuss them adopting @xmath102 in order to look at the difference between the even - even and the odd-@xmath162 cases .
figure [ fig5 ] shows dependence on @xmath0 at @xmath105 0.25 mev with keeping @xmath102 and @xmath163 0.3 mev constant .
figure [ fig5](a ) graphs @xmath55 . in the @xmath94 region ,
the solution is quite similar to the @xmath100hf case . in the @xmath122 region , that for @xmath164 is quite similar again but for @xmath165 its character is completely different . in this region
the presented solution is the lowest in energy and becoming collective gradually as @xmath0 decreases .
the largeness of @xmath55 corresponds to that of @xmath166 in fig.[fig5](b ) .
comparison of figs.[fig5](c ) and [ fig1](c ) certifies that the alignment of the @xmath1 quasiparticle(s ) is almost complete for @xmath94 whereas less for @xmath122 .
this produces quantitative even - odd differences as explained below .
but for @xmath160lu .
[ fig5],title="fig:",width=264 ] but for @xmath160lu .
[ fig5],title="fig:",width=264 ] but for @xmath160lu .
[ fig5],title="fig:",width=264 ] but for @xmath160lu .
[ fig5],title="fig:",width=264 ] having confirmed that these features are independent of @xmath21 and @xmath167 except that we did not obtain any low - lying solutions for @xmath168 in the small-@xmath21 cases , we look into underlying unperturbed 2qp energies to see the even - odd difference . in fig.[fig6 ] we present the energies of the lowest @xmath169 states which represent the biggest difference . in the yrast @xmath106 configuration ,
@xmath170 and @xmath171 in the usual notation are occupied in the even-@xmath162 case , the lowest 2qp state of signature @xmath36 with respect to this is @xmath172 ( where @xmath173 denotes the conjugate state , see eq.([exch ] ) ) . in the odd-@xmath162 case in which @xmath170 is occupied ,
the lowest one is @xmath174 .
since both @xmath175 and @xmath176 decrease as @xmath0 decreases , this 2qp state becomes the dominant component in the lowest - energy rpa solution .
note here that the sum @xmath177 corresponds to the signature splitting between @xmath170 and @xmath171 when they are seen from the usual even - even vacuum .
since both @xmath171 and @xmath178 are of @xmath179 character , the resulting rpa solution can not have the @xmath142 collectivity as shown in fig.[fig5](d ) . according to the relation @xcite , @xmath180^{-1 } , \ ] ] @xmath181 in fig.[fig5](b ) becomes small for @xmath165 .
these discussion serves to exclude the possibility of @xmath88 for the tsds that support collective wobbling excitations in the odd-@xmath162 cases , whereas the even - odd difference in @xmath94 is merely quantitative . two quasiparticle states in @xmath100hf and @xmath160lu , calculated at the same time as figs.[fig1 ] and [ fig5 ] , respectively .
[ fig6],width=264 ] the analyses above indicate that the chosen mean - field parameters are reasonable , and therefore we proceed to study @xmath29-dependence with keeping these parameters constant .
figure [ fig7 ] shows the result for @xmath100hf .
these figures indicate again the @xmath106 alignment that makes @xmath52 larger than @xmath181 is indispensable for the formation of the wobbling excitation . at around @xmath105 0.45
mev the @xmath152 alignment occurs .
in contrast to the low - frequency case reported in fig.[fig3 ] , in the present case its effect on @xmath55 is visible as a small bump .
although the character of the observed excited tsd bands has not been resolved , some anomaly is seen at around this @xmath29 in one of them @xcite .
we suggest this is related to the @xmath152 alignment since this is the only alignable orbital in this frequency region of this shape .
however we note that in @xmath160lu an interaction with a normal deformed state at around this frequency is discussed in ref.@xcite .
we performed calculations also for @xmath183 . in that case , however , wobbling excitation exists only at small @xmath29 because @xmath166 is small as seen from fig.[fig1](b ) .
hf , calculated with @xmath102 , @xmath103 and @xmath104 0.3 mev .
[ fig7],title="fig:",width=264 ] hf , calculated with @xmath102 , @xmath103 and @xmath104 0.3 mev .
[ fig7],title="fig:",width=264 ] hf , calculated with @xmath102 , @xmath103 and @xmath104 0.3 mev .
[ fig7],title="fig:",width=264 ] very recently tsd bands were observed in another even - even nucleus , @xmath182hf @xcite .
it is not trivial if a similar band structure is observed in the nucleus with six neutrons more since the existence of the tsd states depends on the shell gap .
multiple tsd bands were observed but connecting @xmath0-rays have not been resolved also in this nucleus .
we performed a calculation adopting @xmath184 and @xmath185 suggested in ref.@xcite and @xmath104 0.3 mev .
the result is presented in fig.[fig8 ] .
the most striking difference from the case of @xmath100hf above is that @xmath55 decreases steadily as @xmath29 increases after the @xmath106 alignment is completed .
this is because the @xmath152 alignment that causes the small bump in the @xmath100hf case shifts to very low @xmath29 due to the larger neutron number .
hf , calculated with @xmath184 , @xmath185 and @xmath104 0.3 mev .
[ fig8],width=264 ] the wobbling excitation was first observed experimentally in @xmath14lu @xcite , later it was also observed in @xmath186lu @xcite and @xmath160lu @xcite .
the characteristic features common to these isotopes are 1 ) @xmath55 decreases as @xmath29 increases contrary to the consequence of calculations adopting constant moments of inertia , and 2 ) @xmath187 is large typically around 0.2 . here
we concentrate on the isotone of @xmath100hf discussed above , that is , @xmath160lu in order to see the even - odd difference .
a comparison of figs.[fig7 ] and [ fig9 ] proves that all the differences are due to the fact that the number of the aligned @xmath1 quasiparticle is less by one : 1 ) the @xmath106 alignment at around @xmath131 = 0.2 mev is absent , and 2 ) the @xmath188 crossing occurs at around @xmath131 = 0.55 mev , which is proper to the @xmath189 configuration .
figure [ fig9](a ) shows that our calculation does not reproduce the data , although in each frequency range in which the configuration is the same @xmath55 decreases at high @xmath29 as in the cases of the even - even nuclei presented above .
this result might indicate that there is room for improving the mean field .
the @xmath52 in fig.[fig9](c ) is larger than the experimentally deduced value by about 20 30 % .
this is due to the spurious velocity dependence of the nilsson potential mentioned in sec.[sectwob ] .
but for @xmath160lu .
experimental values taken from ref.@xcite are also included in ( a ) .
[ fig9],title="fig:",width=264 ] but for @xmath160lu .
experimental values taken from ref.@xcite are also included in ( a ) .
[ fig9],title="fig:",width=264 ] but for @xmath160lu .
experimental values taken from ref.@xcite are also included in ( a ) .
[ fig9],title="fig:",width=264 ] compared to the excitation energy , the interband @xmath83 values relative to the in - band ones have been measured in only few cases . in fig.[fig10 ] , we report calculated @xmath83 ratios for @xmath190 ( wobbling on yrast tsd ) @xmath191 @xmath192 ( yrast tsd ) transitions in @xmath100hf and @xmath160lu .
the measured ones are also included for the latter . the first point is the magnitude of the larger ( @xmath193 ) ones .
apparently , the calculated @xmath83 values are smaller by factor 2 3 .
the measured interband @xmath83 values amount almost to the macroscopic rotor value . in the rpa calculations , as summarized in sect.[sectwob ] , the @xmath83 value is reduced by a factor @xmath194 ( see eq .
( [ be2wob ] ) ) : only the case with the full - strength @xmath195 the rotor value is recovered .
although the obtained rpa wobbling solutions are extremely collective in comparison with the usual low - lying collective vibrations , like the @xmath119- or @xmath0-vibrations , for which typically @xmath196 0.3 0.4 , this factor is still @xmath95 0.6 0.8 .
this is the main reason why the calculated @xmath83 values are a factor 2 3 off the measured ones . as is well - known
, giant resonances also carry considerable amount of quadrupole strengths , so it seems difficult for the microscopic correction factor @xmath194 to be unity ; it is not impossible , however , because the `` sum rule '' discussed in sect.[sectwob ] is not the sum of positive - definite terms . in the rpa formalism ,
the reduction factor @xmath194 for the @xmath83 value , eq .
( [ be2wob ] ) , comes from the fact that the wobbling motion is composed of the coherent motion of two - quasiparticles , and reflects the microscopic structure of collective rpa solutions .
the measurement that the @xmath83 value suffers almost no reduction may be a challenge to the microscopic rpa theory in the case of the wobbling motion .
calculated @xmath83 ratios for @xmath182hf are slightly smaller than those for @xmath100hf in fig.[fig10](a ) .
transition rates for @xmath190 ( wobbling on yrast tsd ) @xmath191 @xmath197 ( yrast tsd ) transitions in ( a ) @xmath100hf and ( b ) @xmath160lu .
the latter is presented as functions of @xmath198 spin @xmath190 , while the former is presented as functions of the rotational frequency since experimental spin assignment has not been done for @xmath100hf .
the rotational - frequency range corresponding to ( b ) is very narrow in comparison to ( a ) .
interband transition rates are divided by the in - band ones .
experimental values @xcite are also shown in ( b ) . noting that , for @xmath160lu , the states @xmath199 ( tsd1 ) are slightly higher in energy than @xmath190 ( tsd2 ) at @xmath200@xmath201 and @xmath202 holds at high spins , we plotted those for @xmath203 at the places with the abscissae @xmath199 in order to show clearly their characteristic staggering behavior .
[ fig10],title="fig:",width=264 ] transition rates for @xmath190 ( wobbling on yrast tsd ) @xmath191 @xmath197 ( yrast tsd ) transitions in ( a ) @xmath100hf and ( b ) @xmath160lu .
the latter is presented as functions of @xmath198 spin @xmath190 , while the former is presented as functions of the rotational frequency since experimental spin assignment has not been done for @xmath100hf .
the rotational - frequency range corresponding to ( b ) is very narrow in comparison to ( a ) .
interband transition rates are divided by the in - band ones .
experimental values @xcite are also shown in ( b ) .
noting that , for @xmath160lu , the states @xmath199 ( tsd1 ) are slightly higher in energy than @xmath190 ( tsd2 ) at @xmath200@xmath201 and @xmath202 holds at high spins , we plotted those for @xmath203 at the places with the abscissae @xmath199 in order to show clearly their characteristic staggering behavior .
[ fig10],title="fig:",width=264 ] the second point is the staggering , that is , the difference between @xmath204 .
we clarified @xcite its unique correspondence to the sign of @xmath0 as mentioned in subsect.[sectlow ] ; that holds for both even - even and odd-@xmath4 systems .
recently this staggering was discussed from a different point of view @xcite ; but it looks to apply only to @xmath124 cases .
the nuclear wobbling motion , which is a firm evidence of stable triaxial deformations , was identified experimentally in the triaxial superdeformed odd-@xmath4 lu isotopes . in principle , wobbling excitation is possible both in @xmath123 and @xmath124 nuclei .
every information , theoretical and experimental , suggests @xmath123 for these bands . according to the wobbling frequency formula @xcite , c.f .
eq.([disp ] ) , its excitation in nuclei rotating principally about the @xmath26 axis requires @xmath205 , although irrotational - like model moments of inertia give @xmath120 for @xmath123 .
to solve this puzzle , we studied the nuclear wobbling motion , in particular , the three moments of inertia associated with it in terms of the cranked shell model plus random phase approximation .
this makes it possible to calculate the moments of inertia of the whole system including the effect of aligned quasiparticle(s ) .
the results indicate that the @xmath0-dependence of the calculated moment of inertia is basically irrotational - like ( @xmath206 for @xmath207 ) if aligned quasiparticle(s ) ( @xmath1 in the present case ) does not exist .
but once it is excited , it produces an additional contribution , @xmath208 , and consequently can lead to @xmath121 .
this is the very reason why wobbling excitation exists in @xmath123 nuclei . in this sense ,
the wobbling motion is a collective motion that is sensitive to the single - particle alignments .
the resulting moment of inertia for @xmath209 resembles the @xmath0-reversed one , i.e. , the irrotational moment of inertia but with @xmath52 and @xmath93 being interchanged . that for @xmath210 , where single - particle angular momenta dominate , is rigid - body - like .
that for @xmath124 is irrotational - like except for odd-@xmath4 nuclei with @xmath165 where a specific 2qp state determines the lowest rpa solution .
having studied qualitative features of the three moments of inertia at a low rotational frequency , we calculated wobbling bands up to high @xmath29 .
experimentally they were confirmed only in odd-@xmath4 lu isotopes as mentioned above
. the most characteristic feature of the data is that @xmath55 decreases as @xmath29 increases .
this obviously excludes constant moments of inertia . in our calculation three moments of inertia are automatically @xmath29-dependent even when mean field parameters are fixed constant .
it should be stressed that the wobbling - like solution in our rpa calculations is insensitive to the mean - field parameters , especially to the pairing gaps , as is shown in subsect.[sectlowhf ] .
this distinguishes the wobbling - like solution from the usual collective vibrations , which are sensitive to the pairing correlations .
thus , our microscopic rpa calculation confirms that the observed band is associated with a new type of collective excitation , although comparisons to the observed excitation energy indicate that there is room for improving the calculation . as for the interband transition rates , our calculation accounted for only about one half or less of the measured ones , even though the wobbling - like solution is extremely collective compared to the usual vibrational modes
this issue is independent of the details of choosing parameters .
this confronts microscopic theories with a big challenge .
we thank g. b. hagemann for providing us with some experimental information prior to publication .
this work was supported in part by the grant - in - aid for scientific research from the japan ministry of education , science and culture ( nos .
13640281 and 14540269 ) . | the wobbling motion excited on triaxial superdeformed nuclei is studied in terms of the cranked shell model plus random phase approximation .
firstly , by calculating at a low rotational frequency the @xmath0-dependence of the three moments of inertia associated with the wobbling motion , the mechanism of the appearance of the wobbling motion in positive-@xmath0 nuclei is clarified theoretically the rotational alignment of the @xmath1 quasiparticle(s ) is the essential condition .
this indicates that the wobbling motion is a collective motion that is sensitive to the single - particle alignment .
secondly , we prove that the observed unexpected rotational - frequency dependence of the wobbling frequency is an outcome of the rotational - frequency dependent dynamical moments of inertia . |
we consider a smooth bounded and simply connected domain @xmath0 in @xmath1 , with @xmath2 or @xmath3 . inside this domain ,
an incompressible viscous fluid evolves under the navier - stokes equations .
we will name @xmath4 its velocity field and @xmath5 the associated pressure .
the equations read : @xmath6 let us emphasize that the fluid density and the viscosity coefficient are set equal to one for the sake of clarity .
for an impermeable wall , it is natural to prescribe the condition @xmath7 on @xmath8 , where @xmath9 denotes the outward pointing normal to the domain , which means that the fluid can not escape the domain and that there is no cavitation at the boundary . indeed in the case of a perfect fluid , driven by the euler equations rather than the navier - stokes ,
such a condition is sufficient to have existence and uniqueness to the cauchy problem in various appropriate functional settings . for the case here of the navier - stokes equations ,
an extra condition has to be added .
the two following propositions are the most used ( in complement to the previous condition ) : * the no - slip condition @xmath10_{\textrm{tan } } } = 0 $ ] ( dating back to stokes in 1851 ) , where @xmath11_{\textrm{tan}}}:= u - ( u \cdot n ) n\ ] ] denotes the tangential part of the vector field @xmath4 . *
the slip - with - friction condition @xmath12 , where @xmath13_{\textrm{tan } } } \quad \text{with } \quad d(u ) : = \left ( \frac{1}{2}\left(\partial_i u_j + \partial_j u_i\right ) \right)_{1 \leqslant i , j \leqslant d}\ ] ] the rate of strain tensor ( or shear stress ) and @xmath14 is a real constant coefficient for simplicity is a smooth matrix - valued function . ] .
this condition dates back to navier in 1833 ( see @xcite ) .
this coefficient describes the friction near the boundary .
let us observe that , formally , when @xmath15 , the navier condition reduces to the usual no - slip condition .
let us recall the following result , where @xmath16 denotes the closure in @xmath17 of smooth divergence free vector fields which are tangent to @xmath8 .
[ th - leray ] let @xmath18 .
then there exists a global weak solution @xmath4 associated with the initial data @xmath19 .
this result dates back to the pioneering work @xcite by leray where it is proved that @xmath20 .
moreover , leray proved the following partial regularity property : for almost every @xmath21 in @xmath22 , @xmath23 is @xmath24 .
even though leray s paper tackled the case of the no - slip condition , this result can be adapted almost right away to the case of the navier slip - with - friction condition ( see ( * ? ? ?
* section 3 ) ) .
we now assume that we are able to act on a non - empty open part @xmath25 of the full boundary @xmath26 . in particular we may let some fluid enter into the domain and the same volume of fluid go out of the domain ( recall that the fluid is incompressible ) .
then the setting we have in mind now is the following ( see figure [ figure : control_setting ] ) .
* on the part @xmath27 , some boundary conditions are prescribed , either the no - slip condition @xmath28 or the navier condition @xmath7 and @xmath29 ( that is without any source terms or ability to modify the slip coefficient @xmath30 which is assumed to be given once and for all ) . * on the part @xmath25 , we are free to choose a boundary condition which is relevant for some purpose . the control problem ] more precisely we have in mind to drive the system from an arbitrary initial data to some given state at some given time .
the following goal , first suggested by jacques - louis lions in the late 80 s ( cf . for instance @xcite ) tackles the case where the target is the rest state .
[ lbop]*open problem ( op)*. for any @xmath31 and @xmath19 in @xmath32 , does there exist a solution to the navier - stokes system with @xmath33 such that @xmath34 ? above the navier - stokes system to which refers
is constituted of the incompressible navier - stokes equations in @xmath35 and of the no - slip condition or the navier condition on @xmath27 , but without any boundary condition prescribed on the controlled part @xmath25 of the boundary .
such a system is therefore under - determined so that uniqueness of a solution is not expected ( even in the 2d case for which uniqueness of leray solutions is known in the uncontrolled setting corresponding to the case where @xmath36 ) . indeed in the formulation above the control
is implicit : a relevant condition to prescribe as a control on @xmath37 can be recovered by taking the trace on @xmath25 of a convenient solution to the under - determined system .
observe that there is no restriction regarding the sizes neither of the time @xmath31 nor of the initial data @xmath19 in @xmath32 . in the terminology of control theory a positive result to this question amounts to the small - time global controllability of the navier - stokes , or more precisely the _ small - time global exact null controllability _ since the target in is the rest state and has to be reached exactly .
in lions original question , the boundary condition on the uncontrolled part @xmath27 of the boundary is the no - slip boundary condition .
our goal here is to present the following result establishing a positive answer to in the case where some navier conditions are prescribed on @xmath27 .
[ main ] let @xmath31 and @xmath18 .
there exists a weak solution @xmath4 to @xmath38 satisfying @xmath39 and @xmath40 .
theorem [ main ] does not require any condition on the coefficient @xmath30 appearing in the definition of @xmath41 .
indeed , observe that there is no asymptotic parameter in the statement above .
still the next lines about the proof of it will be full of @xmath42 .
let us also mention that the results in @xcite are more general , in particular they prove that one may intercept at time @xmath43 any smooth uncontrolled solution to the navier - stokes system with navier condition on the full boundary @xmath8 .
when jacques - louis lions formulated it in the late 80 s , was pretty impressive since the answer was not even known in the case of the heat equation . for this equation
the first key breakthroughs were obtained by @xcite and @xcite thanks to carleman inequalities respectively associated with parabolic and elliptic second order operators .
the latter has been then extended to the stokes equations and later on to the navier - stokes equations in the case of small initial data by imanuvilov in @xcite .
the smallness assumption implies that the quadratic convective term may be seen as a perturbation term so that the result can be obtained from the controllability of the stokes equations by a fixed point strategy .
this result has since been improved in @xcite by fernndez - cara , guerrero , imanuvilov and puel .
all these works deal with the case of the no - slip boundary condition . for navier slip - with - friction boundary conditions ,
let us mention @xcite and @xcite which prove in particular local null controllability when the initial data is small .
the case of large initial data was first tackled in @xcite , where the first author proves a small - time global result in a 2d setting with navier boundary conditions : the smallness obtained within the inside of the domain is good , but the estimates up to the boundary are not sufficient to conclude using a known local result .
in fact , when there is no boundary , the first author and fursikov prove in @xcite a small - time global exact null controllability result ( in this setting , the control is a source term located in a small subset of the domain ) thanks to the return method and to the global controllability of the incompressible euler equations ( for large smooth initial data ) .
likewise , in @xcite , fursikov and imanuvilov prove small - time global exact null controllability when the control is supported on the whole boundary ( i.e. @xmath44 ) . in @xcite
, chapouly obtains global exact null controllability for navier - stokes in a 2d rectangular domain under lions boundary condition ( corresponding to the case where @xmath45 in the navier condition ) on the uncontrolled part of boundary . still the approaches used in the aforementioned papers failed to deal with the viscous boundary layers appearing near the uncontrolled part of the boundary .
this is precisely the goal of this paper to promote the _ well - prepared dissipation method _ in order to obtain some controllability results despite the presence of boundary layers .
this method was first introduced in @xcite by the second author in order to deduce a controllability result for the 1d burgers equation .
the extension of this method to the navier - stokes equations will be crucial in our proof of theorem [ main ] .
in particular here the method will be implemented thanks to a multi - scale expansion describing the boundary layer occurring in the vanishing viscosity limit of the navier - stokes equations .
the application of the method is presented in section [ wpdm ] .
next sections are devoted to the scheme of proof of theorem [ main ] .
we will try to highlight a few key ingredients whereas some technical difficulties will be omitted on purpose for the sake of clarity .
we refer to @xcite for a complete proof .
let us also mention here that we are not going to really use a control all the time in the sense that it will be relevant on some time intervals to choose as boundary condition on @xmath25 the same navier condition than on @xmath27 so that the system then coincides with the uncontrolled one for which @xmath36 .
in this section we are going to prove that it is sufficient to have the existence of a solution starting from an arbitrary smooth initial data and reaching a state close to zero in @xmath46 , in any positive time in order to conclude the proof of theorem [ main ] . indeed according to leray s partial regularity result hinted above ( cf .
below theorem [ th - leray ] ) , there exists @xmath47 in @xmath48 such that @xmath49 is @xmath24 .
let us assume that we are able to prove the existence of a solution starting from @xmath49 at time @xmath47 and reaching , say at time @xmath50 , a state close enough to zero in @xmath46 such that the local controllability results mentioned above can be applied but leray s partial regularization of the uncontrolled navier - stokes equations can be used again in order to glue these steps .
here we have to pay attention to the preservation of the smallness assumption in this regularization argument ( cf .
@xcite for more ) . ] on the remaining time interval @xmath51 .
then the concatenation of this three steps yields theorem [ main ] ( see figure [ figure : graph1 ] ) .
our task is therefore only to obtain approximate null controllability from a smooth initial data on the intermediate time interval @xmath52 . in order to simplify the notations
let us pretend that this interval is @xmath53 in the next sections . on the other hand we will denote @xmath54 the initial data , which is smooth , for this new problem , in order to distinguish it from the original initial data @xmath19 which was only assumed to be in @xmath32 .
reduction to a global approximate controllability problem ]
in order to take profit of the nonlinearity at our advantage we aim at reaching approximatively zero thanks to a control which is fast and furious in the sense that its amplitude and duration are scaled with respect to a small positive parameter @xmath55 which is introduced by force and will be ultimately taken small enough .
indeed we look for a solution @xmath4 to of the form @xmath56 having in mind to look for a family of functions @xmath57 with , typically , variations of order @xmath58 on time interval of order @xmath58 .
this means for the original searched solution @xmath4 having fast transitions on time interval of order @xmath59 with furious amplitudes of order @xmath60 .
the underlying idea is to start with the ambitious idea to try to control the system during the shorter time interval @xmath61 $ ] with forcing the system to evolve in a high reynolds regime .
regarding the pressure @xmath5 associated with the original solution @xmath4 , having in mind bernoulli s principle which associates the pressure with the square of the velocity , we look for an ansatz of the form @xmath62 this translates the original system in a new system with four main changes : a. the new unknowns @xmath57 and @xmath63 satisfy the navier - stokes equations with a small viscosity coefficient @xmath55 .
b. the initial data for @xmath57 is now small equal to @xmath64 . c. the time interval is now @xmath65 so that we have to investigate the large time behaviour of the system . in particular , although the initial data is small , nonlinearities will matter .
+ the system for @xmath66 therefore reads : @xmath67 d. last change but not least , the rest state is now targeted ( at the final time @xmath68 ) with more precision .
indeed we now plan to prove that there exists a solution @xmath57 to the underdetermined system such that @xmath69 in order to deduce from that there exists a solution @xmath4 to such that @xmath70 . in particular , choosing @xmath55 small enough allows to reach a state arbitrarily close to @xmath71 in @xmath72 .
this will provide the approximate controllability result mentioned in the previous section .
when @xmath55 is small , it is expected that the analysis of the system may be built on the small - time global exact controllability of euler equations .
we therefore consider the following counterpart of the system where the viscosity term has been dropped out : @xmath73 as the initial data is of order @xmath74 in @xmath75 it is natural to for a solution @xmath76 to which is , at least for times of order @xmath58 , of the form : @xmath77 plugging expansions into and grouping terms of order @xmath78 yields : @xmath79 by elementary combinations of the equations we observe that the system does not admit any solution reaching exactly @xmath71 unless the initial data @xmath54 is the gradient of a harmonic function , which is not the case in general .
system suffers from a lack of controllability which will prevent from using it for our purposes . in order to overcome this difficulty we are going to use the return method first introduced by the first author in @xcite
this method takes profit of the nonlinearity thanks to an auxiliary controlled solution to the euler system .
indeed , instead expansions , we will rather look for some asymptotic expansions of the form : @xmath80 where the extra - term @xmath81 is introduced in order to help to control @xmath82 .
of course @xmath81 has to be solution to the euler system in order to cancel out the terms of order @xmath58 which appear when plugging the expansions into the first three equations of .
moreover the last equation yields the initial data @xmath83 in @xmath35 .
the interest is that the equations obtained by gathering the terms of order @xmath74 are now : @xmath84 this is the linearisation of the euler equations around @xmath85 ( rather than around @xmath71 like in ) .
we may now rely on the transport by @xmath85 ( see figure [ figure : control_euler ] ) in order to drive @xmath86 from @xmath87 to @xmath71 , see the second term in the first equation of . more precisely we want to use the transport by @xmath85 in order to flush @xmath86 out of the domain .
of course the system has , in addition to the transport aspect , non local features due to the incompressibility condition .
still , reasoning on the vorticity of @xmath86 , we obtain that if the fluid particles are flushed outside of the physical domain within a time interval of order @xmath58 , say @xmath88 $ ] ( that is if any fluid particle initially at @xmath89 in @xmath0 moves with the flow associated with @xmath85 up to some time @xmath90 for which it reaches @xmath25 with a positive velocity ) , then @xmath86 can be set equal to @xmath71 .
observe that this requires a time interval far smaller than the allotted one which is @xmath91 $ ] .
inviscid flushing ] on the other hand this auxiliary field @xmath85 also has to vanish at the final time . in order to construct such a field
, a crucial observation is that the potential flows as solutions to the euler system enjoy a lot of freedom regarding their behaviour in time . indeed if a scalar function @xmath92 satisfies @xmath93 then for any function @xmath94 , the vector field @xmath95 satisfies the first three equations of for an appropriate pressure .
in particular it is possible to choose a nonzero function @xmath96 satisfying @xmath97 , so that this process leads to a field @xmath85 starting from zero initial data and which vanishes at time @xmath43 .
moreover the set of the scalar functions satisfying the underdetermined neumann problem is rich enough to provide , by an appropriate gluing strategy , vector fields flushing the whole domain on the time interval @xmath88 $ ] .
[ lemma.euler ] there exists a smooth solution @xmath98 to @xmath99 such that the smooth solution @xmath100 to system satisfies @xmath101 in @xmath35 .
this lemma is proved on the one hand by the first author in the papers @xcite and @xcite respectively for 2d simply connected domains and for general 2d domains when @xmath25 intersects all connected components of @xmath26 , and on the other hand by glass in @xcite and @xcite for the corresponding cases in 3d . in the sequel ,
when we need it , we will implicitly extend the previous fields @xmath85 and @xmath86 by zero after @xmath43 .
the difficulty comes from the fact that the euler equation , which models the behavior of a perfect fluid , not subject to friction , is only associated with the boundary condition @xmath102 for an impermeable wall and does not satisfy in general the navier slip - with - friction boundary condition on @xmath27 .
an accurate description of a solution of the navier - stokes equation near @xmath27 , even for a small viscosity , has therefore to use an expansion where a corrector is added to a solution to the euler equation . in the uncontrolled setting for which @xmath36 the description of the behavior of the navier - stokes equation under the navier slip - with - friction condition in the vanishing viscosity limit was performed by iftimie and the third author in @xcite thanks to a multiscale asymptotic expansion involving a boundary layer term @xmath103 of amplitude @xmath104 and of thickness @xmath104 for a vanishing viscosity @xmath55 .
let us first briefly recall this result which will be extended in the sequel to the controlled case .
let us use here temporarily again the notation @xmath85 for a smooth solution to the euler equations on the time interval @xmath88 $ ] with the impermeability condition @xmath105 on the full boundary @xmath26 .
the boundary layer corrector will involve an extra variable describing the fast variations of the fluid velocity in the normal direction near the boundary and will be given as a solution to an initial boundary value problem with a boundary condition with respect to this extra variable .
we introduce a smooth function @xmath106 such that @xmath107 on @xmath8 , @xmath108 in @xmath0 and @xmath109 outside of @xmath110 .
moreover , we assume that @xmath111 in a small neighborhood of @xmath8 . hence , the normal @xmath9 can be computed as @xmath112 close to the boundary and extended smoothly within the full domain @xmath0 .
the notation @xmath113_{\textrm{tan}}}$ ] , introduced in , is extended accordingly .
we also introduce the following definitions : @xmath114 where @xmath115 is a smooth cut - off function satisfying @xmath116 on @xmath8 . even though @xmath117 vanishes on @xmath118 ,
@xmath119 is not singular near the boundary because of the impermeability condition @xmath120 .
indeed since @xmath85 is smooth , a taylor expansion proves that @xmath119 is smooth in @xmath110 .
the boundary layer corrector will be described by a smooth vector field @xmath103 expressed in terms both of the slow space variable @xmath121 and a fast scalar variable @xmath122 , where @xmath123 satisfies the equation : @xmath124_{\textrm{tan } } } + u^0_\flat z \partial_z v - \partial_{zz } v = 0,\ ] ] for @xmath89 in @xmath110 and @xmath125 in @xmath126 , with the following boundary condition at @xmath127 : @xmath128 we refer to ( * ? ? ?
* section 2 ) for a detailed heuristic of the equations and .
let us only mention here that these equations are obtained by plugging @xmath129 into the first and fourth equations of and keeping the terms of higher order ( taking into account that @xmath85 satisfies the euler equations ) .
indeed the pressure @xmath130 has to be expanded as well , into the sum of the euler pressure and of a boundary layer term but the latter can be eliminated from the resulting equation by distinguishing the normal and tangential parts .
thus this pressure boundary layer term acts as a projection on the convective terms and this is why the second term in is only tangential .
the cauchy problem associated with and is well - posed in sobolev spaces .
moreover for any @xmath131 , @xmath132 and @xmath133 , we have @xmath134 it is easy to check that the solution inherits this condition from the initial and boundary data .
this orthogonality property is the reason why equation is linear .
indeed , the quadratic term @xmath135 should have been taken into account if it did not vanish .
thanks to the cut - off function @xmath115 , satisfying @xmath116 on @xmath8 , @xmath103 is compactly supported in @xmath89 near @xmath8 , while ensuring that @xmath103 compensates the navier slip - with - friction boundary trace of @xmath85 .
then it is proved in @xcite that the leray solutions @xmath136 to the navier - stokes equation can be described by the following expansion in @xmath137 : @xmath138 let us highlight that this expansion holds up to any time @xmath31 for which @xmath85 is a smooth solution to the euler equations on the time interval @xmath88 $ ] . on the other hand this analysis fails to describe the vanishing viscosity limit of the navier - stokes equation for large times of order @xmath139 , even in the case where the euler solution stays smooth for all times .
now going back to the controlled setting for which @xmath140 we expect to be able to describe the behavior of the navier - stokes equation near the uncontrolled part @xmath27 of the boundary in the vanishing viscosity limit thanks to a similar expansion . indeed
since we aim at finding a navier - stokes solution satisfying we consider the following refined expansion : @xmath141 where @xmath85 and @xmath86 are as lemma [ lemma.euler ] and the vector field @xmath142 is wished to be @xmath143 at time @xmath144 .
if so , and since the fields @xmath145 and @xmath146 are zero after the time @xmath43 , the leading part of the the expansion after @xmath43 is given by the second term in the right hand side and we therefore must understand the large time behavior of this boundary layer . for @xmath147 , the equations and reduce to @xmath148 where the slow variables @xmath149 play the role of parameters through the `` initial '' data @xmath150 .
this heat system dissipates towards the null equilibrium state .
unfortunately the natural decay ( that is without any assumption on @xmath151 ) at the final time @xmath152 only yields @xmath153 which is not sufficient in view of the wished estimate .
physically , this is due to the fact that the average of @xmath103 is preserved under its evolution by equation and to the fact that the energy contained by low frequency modes decays slowly .
in order to overcome the previous difficulty we are going to use the _ well - prepared dissipation method _ first introduced in @xcite by the second author in order to obtain a new controllability result of the 1d burgers equation in the presence of a boundary layer . we will here adjust the method to the boundary layers associated with the navier conditions in the vanishing viscosity limit of the navier - stokes equations .
the idea is to design a control strategy in order to enhance the natural dissipation of the boundary layer after the time @xmath43 .
our strategy will be to guarantee that @xmath154 satisfies a finite number of vanishing moment conditions for @xmath155 of the form : @xmath156 this will allow to enhance the dissipation and to improve the estimate into @xmath157 actually we aim at constructing the different fields mentioned so far by restriction to the physical domain @xmath0 of solutions to analogous problems in an larger domain @xmath158 extended across @xmath25 ( and @xmath158 can be chosen smooth , bounded and simply connected ) with source terms compactly supported in the added portion of the domain .
this means in particular that we intend to find a solution that we still denote @xmath66 of the following navier - stokes equations : @xmath159 for @xmath89 in @xmath158 where the source term @xmath160 is a vector field supported for @xmath89 in @xmath161 , of the form @xmath162 where @xmath163 and @xmath164 are smooth vector fields used in order to insure lemma [ lemma.euler ] whereas the vector field @xmath165 is devoted to the control of the moments of the boundary layer .
indeed we now aim at obtaining a profile @xmath103 solution to the following equation : @xmath166_{\textrm{tan } } } + u^0_\flat z \partial_z v - \partial_{zz } v = \zeta_v , \ ] ] for @xmath89 in @xmath158 and @xmath125 in @xmath126 .
since the initial boundary value satisfied by @xmath103 is linear , its moments at time @xmath43 ( see the left hand side of ) , can be decomposed as the sum of an addend due to the right hand side of and of an addend due to the outside control ( see the right hand side of ) , which generates some moments outside , and are convected inside the domain by the field @xmath85 , see the second term in .
indeed , according to duhamel s formula , the second addend is given by an integral over the time interval @xmath88 $ ] , which allows to insure the condition for all @xmath89 in @xmath0 .
the kaitenzushi strategy ] let us use here the following metaphor : see the extended domain as a conveyor - belt sushi restaurant , the added part of the extended domain as the kitchen and the moments as the plates ( see figure [ figure : kaitenzushi ] ) . in order to send some plates from the kitchen , without sending the chef into the dining room
, we use the transport by the field @xmath85 as a conveyor belt to serve the wished moments ( compensating what comes from the uncontrolled part of the boundary ) all along the boundary before the end of the service , which corresponds here to the time @xmath43 . in this process
it is crucial to maintain the orthogonality condition ( otherwise the linearity of the equation would dramatically fall down because of the term @xmath135 mentioned above ) .
let us observe that it seems impossible to control completely the boundary layer @xmath103 because the plates need some time to be conveyed from the kitchen and are therefore strongly regularized in @xmath125 ( the equation is parabolic in @xmath125 ) when they are supposed to be compensating what comes instantly from the nonhomogeneous data on the uncontrolled part of the boundary and is therefore far less regularized .
thus a compensation is only possible for a projection on a functional space containing the two types of contributions and we precisely make use of some finite dimensions projections by adjusting a finite number of moments .
going back to the velocity expansion we are led now to the issue of estimating the large time behaviour of the remainder @xmath167 .
the field @xmath167 can be naturally defined as the solution to a navier - stokes type equation of the form : @xmath168 where @xmath169 denotes the pressure associated with the vector field @xmath142 , the notation @xmath170 stands for an amplification operator and @xmath171 for a source term both due to the terms which were omitted in the equations of @xmath85 , @xmath86 and @xmath103 for being of higher order in @xmath55 .
since the field @xmath146 bears the initial data @xmath172 , this remainder starts with a zero initial data ( taking the trace at the initial time of the equality and taking into account that @xmath85 and @xmath103 start with zero initial data ) but is generated by the source term @xmath171 and possibly amplified by mean of the term @xmath173 .
of course the equation is completed with the divergence free condition and some initial and boundary conditions . here
the key points in the large time estimate of @xmath142 are on the one hand that the quadratic nonlinearity in term of @xmath142 which is hidden in the third term of is tamed by a factor @xmath55 , see the velocity expansion , and on the other hand that the effects of both @xmath170 and @xmath171 are tamed by the enhanced dissipation hinted in section [ wpdm ] .
let us refer once more to @xcite for more on the technicalities and only conclude here that the result of an energy estimate is that @xmath174
taking into account that the fields @xmath85 and @xmath86 vanish after @xmath43 , estimates and plugged into expansion yield and therefore conclude the proof of theorem [ main ] thanks to the preliminary reductions performed in sections [ sec - red ] and [ sec - ff ] .
the evolution of the state during the control strategy is pictured in figure [ figure : graph2 ] .
four main steps of the evolution ] the main steps of the proof can be summarized as in table [ table ] .
.[table]main features of the control steps [ cols="^,^,^,^,^ " , ]
let us mention a few questions inspired by this work : 1
. provided a smooth initial data , is there a strong solution to the 3d navier - stokes system reaching zero at time @xmath43 ?
the 2d case follows from theorem [ main ] .
is it possible to deduce from the previous analysis some lagrangian controllability results ? this would extend the results obtained in @xcite , @xcite for the incompressible euler equations and in @xcite for the stationary stokes equation.this issue is actually related to the previous one as lagrangian setting requires enough regularity for the flow to be controlled .
3 . last but not least , is it possible to tackle in the more difficult case of the no - slip boundary condition , at least for some favorable geometric settings ? this is a very challenging open problem because the no - slip boundary condition gives rise to boundary layers that have a larger amplitude than navier slip - with - friction boundary layers .
we refer to the nice recent survey @xcite by maekawa and mazzucato for more on boundary layers in the no - slip case .
the two first authors were partly supported by erc advanced grant 266907 ( cpdenl ) of the 7th research framework programme ( fp7 ) .
the third author thanks the agence nationale de la recherche , project dyficolti , grant anr-13-bs01 - 0003 - 01 , project ifsmacs , grant anr-15-ce40 - 0010 for their financial support and the third author thanks the hospitality of rims during the workshop on mathematical analysis of viscous incompressible fluid " .
jean - michel coron , frdric marbach , franck sueur , small time global exact null controllability of the navier - stokes equation with navier slip - with - friction boundary conditions .
preprint 2016 .
enrique fernndez - cara , sergio guerrero , oleg imanuvilov , and jean - pierre puel .
local exact controllability of the navier - stokes system . , 83(12):15011542 , 2004 .
teodor havrneanu , ctlin popa , and sivaguru sritharan .
exact internal controllability for the two - dimensional navier - stokes equations with the navier slip boundary conditions .
, 55(12):10221028 , 2006 .
andrei v. fursikov , and oleg y. imanuvilov .
controllability of evolution equations . , 34 .
seoul national university , research institute of mathematics , global analysis research center , seoul , 1996 .
iv+163 pp . jacques - louis lions .
exact controllability for distributed systems . some trends and some problems . in _
applied and industrial mathematics ( venice , 1989 ) _ , volume 56 of _ math .
_ , pages 5984 .
kluwer acad .
publ . , dordrecht , 1991 . | in this proceeding we expose a particular case of a recent result obtained in @xcite by the authors regarding the incompressible navier - stokes equations in a smooth bounded and simply connected bounded domain , either in 2d or in 3d , with a navier slip - with - friction boundary condition except on a part of the boundary .
this under - determination encodes that one has control over the remaining part of the boundary .
we prove that for any initial data , for any positive time , there exists a weak leray solution which vanishes at this given time . |
the construction of the model , that follows a hierarchical rule commonly used in deterministic fractals @xcite , is shown in figure 1 .
the network is built in an iterative fashion , each iteration repeating and reusing the elements generated in the previous steps as follows : _ step 0 : _ we start from a single node , that we designate as the _ root _ of the graph . _ step 1 : _ we add two more nodes , and connect each of them to the root .
_ step 2 : _ we add two units of three nodes , each unit identical to the network created in the previous iteration ( step 1 ) , and we connect each of the _ bottom _ nodes ( see figure 1 ) of these two units to the root .
that is , the root will gain four more new links .
_ step 3 : _ we add two units of nine nodes each , identical to the units generated in the previous iteration , and connect all eight bottom nodes of the two new units to the root . these rules can be easily generalized .
indeed , step @xmath4 would involve the following operation : _ step @xmath4 : _ add two units of @xmath5 nodes each , identical to the network created in the previous iteration ( step @xmath6 ) , and connect each of the @xmath7 bottom nodes of these two units to the root of the network .
thanks to its deterministic and discrete nature , the model described above can be solved exactly . to show its scale - free nature , in the following we concentrate on the degree distribution , @xmath8 .
the tail of the degree distribution is determined by the most connected nodes , or hubs .
clearly the biggest hub is the root , and the next two hubs are the roots of the two units added to the network in the last step .
therefore , in order to capture the tail of the distribution , it is sufficient to focus on the hubs . in step
@xmath9 the degree of the most connected hub , the root , is @xmath10 . in the next iteration
two copies of this hub will appear in the two newly added units .
as we iterate further , in the @xmath4th step @xmath11 copies of this hub will be present in the network .
however , the two newly created copies will not increase their degree after further iterations . therefore ,
after @xmath4 iterations there are @xmath12 nodes with degree @xmath10 .
since spaces between degrees grow with increrasing @xmath0 , the exponent of the degree distribution can be calculated using the cumulative degree disctribution .
the tail of the cumulative degree distribution , determined by the hubs , follows @xmath13 thus the degree exponent is @xmath14 the origin of this scaling behavior can be understood by inspecting the model s construction .
indeed , at any moment we have a hierarchy of hubs , highly connected nodes which are a common component of scale - free networks .
the root is always the largest hub .
however , at any step there are two hubs whose connectivity is roughly a half of the root s connectivity , corresponding to the roots of the two units added at step @xmath6 .
there are six even smaller hubs , with connectivity @xmath15 , corresponding to the root of the units added at time @xmath16 , and so on .
this hierarchy of hubs is responsible for the network s scale - free topology .
the model introduced above offers a deterministic construction of a scale - free network .
one of its interesting properties is its apparent self - similarity . on the other hand , while the networks generated at step @xmath4 and @xmath6 have an identical large - scale topology , self - similarity is not complete . the special role played by the root , the major hub in the network , violates complete self - similarity .
indeed , the root keeps a detailed record of the system size through the number of links it has .
such global information is never present in a local element of a fractal , the common example of a self - similar object .
this unique feature of our construction reflects a common property of all scale - free networks , rooted in the nonlocal growth rule that generates scale - free systems . indeed , in stochastic models the probability that a node connects to an existing node in the system contains implicitly the information about the whole system .
while the connectivity of the hub in a stochastic network is not equal to the system size , it typically varies as a simple function of @xmath17 .
similarly , here the largest hub , the root of our construction , keeps track of the system size in a trivial way , as @xmath18 of the nodes are linked to it .
the proposed model generates a network with a fixed @xmath19 degree exponent .
however , one can easily modify the model to change the scaling exponent by varying the number of links connected to the root at each step .
similarly , by definition the model discussed here has a zero clustering coefficient @xcite , as it does not generate triangles of connected nodes .
it is easy to change the rules , without changing the scaling exponent , to obtain a network that displays nonzero clustering .
indeed , a version in which each two copies of the original hub are connected has a finite clustering coefficient .
these and further variants of the model will be discussed in forthcoming publications . a crucial unique property of the model is its hierarchical structure .
it is not clear whether the hierarchy seen in this model is a unique property of its construction , or it is an intrinsic property of all scale - free networks . uncovering the elements of the hierarchy could be an important future task in our understanding of scale - free networks .
the present model explicitly builds such a hierarchical construction into the network , as at each iteration we connect nodes to each other in a hierarchical fashion . in conclusion
, we introduced a simple model that allows us to construct a deterministic scale - free network .
we have shown that the scaling exponent characterizing the tail of the @xmath8 distribution can be calculated analytically .
an important feature of the model is its hierarchical topology , as at each iteration it combines identical elements to generate a larger network .
the method that lead to the construction of the network naturally lends itself to immediate generalizations , leading to structures that are similar in spirit , but rather different in detail , allowing us to generate networks with different scaling exponents and connectivity .
albert , r. , barabsi , a.l .
, rev . of mod .
( 2001 ) ( in press ) , cond - mat/0106096 .
barabsi , a.l . , physics world , vol .
* 14 * , 7 , ( 2001 ) 33 - 38 .
dorogovtsev , s. n. , mendes j. f. f. , adv . in phys .
( 2001 ) ( in press ) , cond - mat/0106144 .
amaral , l. a. n. , a. scala , m. barthlmy and h. e. stanley , proc .
usa * 97 * , ( 2000 ) 11149 .
newman , m. e. j. , proc .
98 ( 2001 ) 404 - 409 .
newman , m. e. j. , j. stat . phys .
* 101 * , ( 2000 ) 819 .
barabsi , a.l . ,
jeong , h. , nda , z. , ravasz , e. , schubert , a. , vicsek , t. , cond - mat/0104162 ( 2001 ) .
faloutsos , m. , p. faloutsos and c. faloutsos , proc .
acm sigcomm , comput .
commun . rev . * 29 * , ( 1999 ) 251 .
albert , r. , jeong , h. , barabsi , a.l .
, nature * 406 * , ( 2000 ) 378 .
albert , r. , jeong , h. , barabsi , a.l .
, nature * 401 * , ( 1999 ) 130 .
jeong , h. , tombor , b. , albert , r. , oltvai , z.n . ,
barabsi , a.l . ,
nature * 407 * , ( 2000 ) 651 - 654 .
jeong , h. , mason , s. , barabsi , a.l .
, oltvai , z.n . ,
nature * 411 * , ( 2001 ) 41 - 42 .
barabsi , a.l . ,
albert , r. , science * 286 * , ( 1998 ) 509 .
barabsi , a.l . ,
albert , r. , jeong , h. , physica a * 272 * , ( 1999 ) 173 .
broder , a. , r. kumar , f. maghoul , p. raghavan , s. rajalopagan , r. stata , a. tomkins and j. wiener , comput .
netw . * 33 * , ( 2000 ) 309 .
liljeros , f. , edling , c.r . ,
amaral , l.a.n . ,
stanley , h.e . ,
aringberg , y. , nature * 411 * , ( 2001 ) 907 - 908 .
sol , r. v. and j. m. montoya , ( 2000 ) cond - mat/0011196 .
albert , r. and a .-
barabsi , phys .
lett . * 85 * , ( 2000 ) 5234 .
dorogovtsev , s. n. , mendes j. f. f. , europhys . lett . * 50 * , ( 2000 ) 1 .
dorogovtsev , s. n. , mendes j. f. f. , phys .
e * 62 * , ( 2000 ) 1842 .
dorogovtsev , s. n. , mendes j. f. f. , europhys .
lett * 52 * , ( 2000 ) 33 .
dorogovtsev , s. n. , mendes j. f. f.and a. n. samukhin , phys .
lett . * 85 * , ( 2000 ) 4633 .
krapivsky , p. l. , redner s. , phys .
e * 63 * , 066123 ( 2001 ) .
p. l. , redner s. , leyvraz , f. , phys .
lett * 85 * , ( 2000 ) 4629 .
krapivsky , p. l. , rodgers , g.j .
, redner , s. , phys .
( to appear ) , ( 2001 ) cond - mat/0012181 .
bianconi , g. , barabsi , a.l .
, europhys . lett .
* 54 * ( 4 ) , 443 - 448 ( 2001 ) .
bianconi , g. , barabsi , a.l .
rev . lett . * 86 * , 5632 ( 2001 ) .
jeong , h. , z. nda and a .-
barabsi , ( 2001 ) cond - mat/0104131 .
pastor - satorras , r. , vzquez a. , vespignani , a. , cond - mat/0105161 .
mandelbrot , b.b . , _ the fractal geometry of nature _ ( freeman , san fransisco ) 1982 .
vicsek , t. , _ fractal growth phenomena _ ( world scientific , singapore , new jersey ) 1989 . | scale - free networks are abundant in nature , describing such diverse systems as the world wide web , the web of human sexual contacts , or the chemical network of a cell .
all models used to generate a scale - free topology are stochastic , that is they create networks in which the nodes appear to be randomly connected to each other . here
we propose a simple model that generates scale - free networks in a deterministic fashion .
we solve exactly the model , showing that the tail of the degree distribution follows a power law .
disordered systems ; networks ; scale - free networks ; scaling networks are ubiquitous in nature and society @xcite , describing various complex systems , such as the society , a network of individuals linked by various social links @xcite ; the internet , a network of routers connected by various physical connections @xcite ; the world wide web , a virtual web of documents connected by uniform resource connectors @xcite or the cell , a network of substrates connected by chemical reactions @xcite . despite their diversity ,
most networks appearing in nature follow universal organizing principles .
in particular , it has been found that many networks of scientific interest are scale - free @xcite , that is , the probability that a randomly selected node has exactly @xmath0 links decays as a power law , following @xmath1 , where @xmath2 is the degree exponent .
the list of documented scale - free networks now include the world wide web @xcite , the internet @xcite , the cell @xcite , the web of human sexual contacts @xcite , the language @xcite , or the web of actors in hollywood @xcite , most of which appear to have degree exponents between two and three .
the @xmath3high@xmath3 interest@xmath3 in@xmath3 understanding@xmath3 the@xmath3 topology@xmath3 of @xmath3complex @xmath3networks@xmath3 has @xmath3resulted@xmath3 in@xmath3 the@xmath3 development@xmath3 of a@xmath3 considerable@xmath3 number @xmath3of @xmath3 network @xmath3models @xcite .
most of these are based on two mechanisms : incremental growth and preferential attachment @xcite .
incremental growth captures the fact that networks are assembled through the addition of new nodes to the system , while preferential attachment encodes the hypothesis that new nodes connect with higher probability to more connected nodes .
both of these mechanisms have been supported by extensive empirical measurements @xcite , indicating that they are simultaneously present in many systems with scale - free network topology .
stochasticity is a common feature of all network models that generate scale - free topologies .
that is , new nodes connect using a probabilistic rule to the nodes already present in the system .
this randomness present in the models , while in line with the major features of networks seen in nature , makes it harder to gain a visual understanding of what makes them scale - free , and how do different nodes relate to each other .
it would therefore be of major theoretical interest to construct models that lead to scale - free networks in a deterministic fashion . here
we present such a simple model , generating a deterministic scale - free network using a hierarchical construction . |
in the study of bivariate polynomials orthogonal on the bi - circle progress has recently been made in understanding these polynomials in the case when the orthogonality measure is purely absolutely continuous with respect to lebesgue measure of the form @xmath0 where @xmath1 is of degree @xmath2 in @xmath3 and @xmath4 in @xmath5 and is stable i.e. is nonzero for @xmath6 and @xmath7 is the normalized lebesgue measure on the torus @xmath8 .
such measures have come to be called bernstein - szeg measures and they played an important role in the extension of the fejr - riesz factorization lemma to two variables @xcite , @xcite , @xcite , @xcite .
in particular in order to determine whether a positive trigonometric polynomial can be factored as a magnitude square of a stable polynomial an important role was played by a bivariate analog of the christoffel - darboux formula .
the derivation of this formula was non trivial even if one begins with the stable polynomial @xmath9 , @xcite , @xcite , @xcite , @xcite .
this formula was shown to be a special case of the formula derived by cole and wermer @xcite through operator theoretic methods . here
we give an alternative derivation of the christoffel - darboux formula beginning with the stable polynomial @xmath9 .
this is accomplished by examining the orthogonality properties of the polynomial @xmath9 in the space @xmath10 .
these orthogonality properties imbue certain subspaces of @xmath10 with many more orthogonality relations than would appear by just examining the defining relations for these spaces .
we proceed as follows . in section 2
we introduce the notation to be used throughout the paper and examine the orthogonality properties of the stable polynomial @xmath9 in the space @xmath10 .
we also list the properties of a sequence of polynomials closely associated with @xmath9 . in section 3
we state , and in section 4 , prove , one of the main results of the paper on the orthogonality of certain subspaces of @xmath10 .
we also establish several follow - up results which are then used in section 5 to derive the christoffel - darboux formula .
the proof is reminiscent of that given in @xcite and @xcite . in section 6 ,
we study connections to the parametric moment problem .
let @xmath11 $ ] be stable with degree @xmath2 in @xmath3 and @xmath4 in @xmath5 .
we will frequently use the following partial order on pairs of integers : @xmath12 the notations @xmath13 refer to the negations of the above partial order . define @xmath14 when we refer to `` orthogonalities
, '' we shall always mean orthogonalities in the inner product @xmath15 of the hilbert space @xmath16 on @xmath8 .
notice that @xmath16 is topologically isomorphic to @xmath17 but we use the different geometry to study @xmath9 .
the polynomial @xmath18 is orthogonal to more monomials than the one variable theory might initially suggest .
more precisely , [ porth ] in @xmath19 , @xmath9 is orthogonal to the set @xmath20 and @xmath21 is orthogonal to the set @xmath22 observe that since @xmath23 is holomorphic in @xmath24 @xmath25 by the mean value property ( either integrating first with respect to @xmath3 or @xmath5 depending on whether @xmath26 or @xmath27 ) . the claim about @xmath21 follows from the observation @xmath28 .
[ orthsetp ] write @xmath29 .
since @xmath1 is stable it follows from the schur - cohn test for stability @xcite that the @xmath30 matrix @xmath31 is positive definite for @xmath32 . here
@xmath33 .
define the following parametrized version of a one variable christoffel - darboux kernel @xmath34 t_m(z)[1,\ldots,\eta^{m-1}]^{\dag}\\\nonumber & = \sum_{j=0}^{m-1 } a_j(z , w)\bar \eta^j,\end{aligned}\ ] ] where @xmath35 are polynomials in @xmath36 , as the following lemma shows in addition to several other important observations .
[ chrisdar ] let @xmath1 be a stable polynomial of degree @xmath37 . then , 1 .
@xmath38 is a polynomial of degree @xmath39 in @xmath36 and a polynomial of degree @xmath40 in @xmath41 .
2 . @xmath42 spans a subspace of dimension @xmath4 as @xmath43 varies over @xmath44 .
3 . @xmath38 is symmetric in the sense that @xmath45 so @xmath46 .
@xmath38 can be written as @xmath47 where @xmath48 are polynomials of degree @xmath49 in @xmath50 .
the numerator of @xmath38 vanishes when @xmath51 , so the factor @xmath52 divides the numerator .
this gives ( 1 ) . for ( 2 ) , when @xmath53 use equation .
since @xmath54 for @xmath32 , @xmath55 spans a set of polynomials of dimension @xmath4 . for ( 3 ) , this is just a computation . for ( 4 ) , observe that ( suppressing the dependence of @xmath56 on @xmath2 and @xmath4 ) , @xmath57
our main goal is to prove that @xmath38 and @xmath58 possess a great many orthogonality relations in @xmath19 .
the orthogonality relations of @xmath38 are depicted in figure [ orthset_l ] .
[ mainprop ] in @xmath19 , each @xmath59 is orthogonal to the set @xmath60 in @xmath19 , @xmath61 is orthogonal to the set @xmath62 note that @xmath63 [ cor : ak ] in @xmath19 , the polynomial @xmath59 is uniquely determined ( up to unimodular multiples ) by the conditions : @xmath64 @xmath65 and @xmath66 ( the last fact follows from proposition [ prop : lnorm ] , which is not currently essential . ) [ akremark ] we emphasize that ( 1 ) each @xmath59 is explicitly given from coefficients of @xmath9 , ( 2 ) each @xmath59 is determined by the orthogonality relations in corollary [ cor : ak ] ( depicted in figure [ aksupporth ] ) , and ( 3 ) each satisfies the additional orthogonality relations from theorem [ mainprop ]
. one useful consequence of this is that the set @xmath67 is dual to the monomials @xmath68 within in the subspace @xmath69 namely , @xmath70 unless @xmath71 and @xmath72 . in particular ,
if @xmath73 , then @xmath74
we begin by writing @xmath75 recall equation and lemma [ chrisdar ] item ( 4 ) . by examining coefficients of @xmath76 in @xmath38 @xmath77 also ,
@xmath78 and @xmath79 have at most degree @xmath80 in @xmath5 . to see this ,
recall equation and observe that @xmath81 which shows that @xmath82 has degree at most @xmath80 in @xmath5 ( i.e. powers of @xmath5 only occur next to greater powers of @xmath43 ) .
the same holds for @xmath83 . by lemma [ porth ]
, @xmath9 is orthogonal to @xmath84 and since @xmath85 has degree at most @xmath2 in @xmath3 and @xmath86 in @xmath5 , @xmath87 also , @xmath88 since the orthogonality relation for @xmath21 ( also from lemma [ porth ] ) is unaffected by multiplication by holomorphic monomials . hence , @xmath89 is orthogonal to the intersection of these sets ; namely , @xmath90 since @xmath91 and since @xmath92 , @xmath93 hence , @xmath59 is orthogonal to the union of the sets in and .
the set in contains @xmath94 and the set in contains @xmath95 .
also , the set in contains @xmath96 while the set in contains @xmath97 . combining all of this we get @xmath98 . finally , @xmath38 is orthogonal to the intersection of @xmath99 .
we now look at the space generated by shifting the @xmath59 s by powers of @xmath3 .
[ gw ] with respect to @xmath100 , @xmath101 and this is orthogonal to the larger set @xmath102 since the @xmath59 are polynomials of degree at most @xmath40 in @xmath5 , it is clear that @xmath103 by theorem [ mainprop ] , the @xmath59 are orthogonal to the spaces @xmath104 and since these spaces are invariant under multiplication by @xmath105 , the polynomials @xmath106 are also orthogonal to these spaces for all @xmath107 .
so , @xmath108 therefore , @xmath109 and this containment must in fact be an equality . indeed , any @xmath110 in @xmath111 which is orthogonal to @xmath112 satisfies @xmath113 for @xmath114 and @xmath115 by remark [ akremark ] and equation .
such an @xmath110 can not also be orthogonal to the space @xmath116 without being identically zero .
define @xmath117 define also the reflection
@xmath118 @xmath119 [ reph ] we have the following orthogonal direct sum decompositions in @xmath120 @xmath121 @xmath122 if @xmath123 is the reproducing kernel for @xmath124 and @xmath125 is the reproducing kernel for @xmath118 , then the reproducing kernel for the spaces in and are given by @xmath126 respectively .
now @xmath124 is an @xmath4 dimensional space of polynomials contained in the space of the previous theorem . in particular ,
@xmath127 and @xmath128 since this space is also @xmath4 dimensional and contains @xmath124 . from this
it is clear that @xmath129 for @xmath26 and we have @xmath130 since shifts of @xmath124 are contained in @xmath131 , we must have @xmath132 which combined with gives .
next , @xmath118 is also @xmath4 dimensional and by is orthogonal to @xmath133 which in particular contains the strip @xmath134 .
so , @xmath135 by dimensional considerations . therefore , @xmath136 the formulas for the reproducing kernels are direct consequences of the orthogonal decompositions ( see @xcite for more on this ) . in @xmath16 the reproducing kernel for @xmath137 is @xmath138 first , @xmath139 is the reproducing kernel for @xmath140 since @xmath141 by the cauchy integral formula . on the other hand , @xmath142 is the reproducing kernel for @xmath143 to see this it is enough to show that @xmath144 is an orthonormal basis for the space . by lemma [ porth ]
, @xmath145 is in the space in for every @xmath146 and it is easy to check that these polynomials form an orthonormal set . we show that their span is dense
. we may write @xmath147 with @xmath148 , since @xmath9 is stable . now , let @xmath110 be in the space in .
if @xmath149 , then since @xmath110 is already orthogonal to the `` lower order terms '' we see that @xmath150 .
inductively , then , we see that assuming @xmath151 for all @xmath152 and @xmath153 but @xmath154 and assuming @xmath155 , we automatically get @xmath156 since @xmath110 will be orthogonal to the lower order terms in @xmath157 . therefore ,
if @xmath110 in is orthogonal to @xmath158 there can be no minimal @xmath159 ( in the partial order on pairs ) such that @xmath110 is not orthogonal to @xmath160 .
in particular , @xmath161 for all @xmath162 and @xmath163 and by @xmath164 , which forces @xmath165 .
so , @xmath166 is an orthonormal basis for the space in while is the reproducing kernel for this space . finally , the reproducing kernel for @xmath167 is the difference of the reproducing kernels we have just calculated .
namely , @xmath138
set @xmath168 and @xmath169 the two variable christoffel - darboux formula is the following . [ cd ] let @xmath9 be a stable polynomial . let @xmath170 be the reproducing kernel for @xmath171 and let @xmath172 be the reproducing kernel for @xmath173 .
then @xmath174 set @xmath175 and notice that @xmath176 and @xmath177 together span @xmath178 .
theorem [ gw ] says @xmath179 which a fortiori implies @xmath180 to see this , suppose @xmath181 .
then , @xmath182 . as @xmath176 and @xmath177 span @xmath178
, such an @xmath110 must be orthogonal to all of @xmath178 and must equal @xmath183 . the reproducing kernel for the space @xmath184 from proposition [ reph ] .
if we interchange the roles of @xmath3 and @xmath5 in proposition [ reph ] we see that @xmath185 is the reproducing kernel for @xmath177 . finally , the reproducing kernel for @xmath186 can be written in two ways .
on the one hand it equals @xmath187 but on the other it equals @xmath188 by the discussion above . equating these formulas and multiplying through by @xmath189 , yields the desired formula .
the above results also shed light on the parametric orthogonal polynomials .
the following proposition shows that the inner products of @xmath190 with respect to @xmath191 for the measures parametrized by @xmath192 @xmath193 are _ trigonometric polynomials _ in @xmath3 .
for @xmath194 the expression @xmath197 is the reproducing kernel / christoffel - darboux kernel for polynomials in @xmath5 of degree at most @xmath40 with respect to the measure @xmath198 .
indeed , this is one of the main consequences of the christoffel - darboux formula in one variable ( see @xcite equation ( 34 ) or @xcite theorem 2.2.7 ) .
it is a general fact about reproducing kernels @xmath199 that @xmath200 using these two observations , follows .
equation follows from matching the coefficients of @xmath201 in .
given @xmath202 defined in equation , set @xmath203 as the determinant of the @xmath204 submatrix of @xmath205 obtained by keeping the first @xmath206 rows and columns and set @xmath207 .
we now perform the lu decomposition of @xmath208 which because it is positive definite does not require any pivoting
. set @xmath209^t = u(\theta)[w^{m-1},\ldots , 1]^t,\ ] ] where @xmath210 is the upper triangular factor obtained from the lu decomposition of @xmath208 without pivoting .
we find : which uniquely specify the polynomials .
the above implies @xmath215 ^ 2}e^{i\theta k } d_{m - j+1}(\theta)\phi^{\theta}_j(e^{i\phi})\overline{\phi^{\theta}_j(e^{i\phi})}\frac{d\theta d\phi}{(2\pi)^2|p_{n , m}(e^{i\theta},e^{i\phi})|^2}=0 , \ k >
n(m - j).\ ] ] from the definition of @xmath208 we see that it is the inverse of the @xmath30 moment matrix associated with @xmath216 .
the first part of the result now follows from the one dimensional theory of polynomials orthogonal on the unit circle .
the second part follows since @xmath217 is polynomial in @xmath3 .
a. grinshpan , d. s. kaliuzhnyi - verbovetskyi , v. vinnikov , and h. j. woerdeman , _ classes of tuples of commuting contractions satisfying the multivariable von neumann inequality _ , j. funct .
57 * ( 2009 ) , 30353054 .
b. simon , _ orthogonal polynomials on the unit circle .
_ , classical theory .
american mathematical society colloquium publications * 54 * , part 1 .
american mathematical society , providence , ri , 2005 . | the orthogonality properties of certain subspaces associated with bivariate bernstein - szeg measures are considered .
it is shown that these spaces satisfy more orthogonality relations than expected from the relations that define them .
the results are used to prove a christoffel - darboux like formula for these measures . |
in the alvarez - macovski@xcite method , the line integrals of the x - ray basis set coefficients are computed from measurements with multiple spectra .
the introduction of photon counting detectors into medical x - ray imaging@xcite gives the possibility of providing the spectral data by pulse height analysis ( pha ) .
these detectors , however , have multiple defects@xcite that affect the information they provide . in some cases , the defects have been found to produce sharply increased noise in the estimates of the line integrals at specific object attenuations@xcite .
the increased noise may indicate potential non - invertibility of the transformation between the spectral measurements and the line integrals .
therefore , it is important to develop mathematical descriptions of the invertibility of the transformation .
this paper is a step towards this mathematical description .
it presents a proof that measurements with two spectra and a photon counting detector with pileup are invertible .
an example of this would be measurements with two x - ray tube voltages .
note that the results of this paper are not applicable to two bin pulse height analysis with pileup case since the effective spectra for that case depend on the object thickness .
for biological materials we can approximate the x - ray attenuation coefficient @xmath0 accurately with a two function basis set@xcite @xmath1 in this equation , @xmath2 are the basis set coefficients and @xmath3 are the basis functions , @xmath4 . as implied by the notation , the coefficients @xmath2
are functions only of the position @xmath5 within the object and the functions @xmath3 depend only on the x - ray energy @xmath6 .
if there is a high atomic number contrast agent , we need to extend the basis set to higher dimensions .
neglecting scatter , the expected value of the number of transmitted photons @xmath7 with an effective measurement spectrum @xmath8 is @xmath9 where the line integral in the exponent is on a line from the x - ray source to the detector . using the decomposition , eq .
[ eq:2-func - decomp ] , the line integral in eq .
[ eq : ik - integral ] is @xmath10 where @xmath11 are the line integrals of the basis set coefficients .
the @xmath12 are summarized as the components of the a - vector , @xmath13 , and the measurements by a vector , @xmath14 , whose components are the photon counts with each effective spectrum . since the body transmission is exponential in @xmath13 , we can approximately linearize the measurements by taking logarithms .
the results is the log measurement vector @xmath15 , where @xmath16 is the expected value of the measurements with no object in the beam and the division means that corresponding members of the vectors are divided .
equations [ eq : ik - integral ] define a relationship between @xmath13 and the expected value of the measurement vector , @xmath17 . the invertibility of this transformation is the subject of this paper .
in a previous paper@xcite , i showed that without pileup the transformation @xmath17 with two spectra and a two function basis set is invertible under quite general conditions .
since i will use parts of the derivation with the pileup case , i will repeat it here . before discussing the situation in general ,
two simple but important cases will be discussed .
the first case is two monoenergetic spectra with energies @xmath18 and @xmath19 . in this case
the expected value of the measurement in eq .
[ eq : ik - integral ] can be linearized by taking the logarithm .
the transformation can be written @xmath20 where @xmath21 is a matrix with coefficients @xmath22\ ] ] the transformation is invertible if the determinant of @xmath21 is not equal to zero : @xmath23 that is if @xmath24 figure [ fig : ratio - of - optimal funcs ] shows a plot of the ratio of the basis set functions versus energy in the medical diagnostic region .
note that the ratio is monotonically increasing .
thus , the condition in equation ( [ eq : f ratios ] ) will be true if the two energies are different .
is satisfied.[fig : ratio - of - optimal funcs ] ] another useful special case is when the object is known to consist of a single material with basis set coefficient vector @xmath25 . in this case , a single photon count measurement suffices to determine the material thickness . with a single material of thickness @xmath26
, the @xmath13 vector is @xmath27 and the expected number of photons is a function of only @xmath26 @xmath28de.\ ] ] the derivative of @xmath29 with respect to @xmath26 is always negative @xmath30de<0 .
\end{array}\label{eq : lambda - prime}\ ] ] because the integrand is always greater than zero . since @xmath31 is monotonically decreasing , it can be inverted to compute @xmath26 .
for the general case , the following theorem is useful ( fulks 1978@xcite page 284 ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let f be a continuously differentiable mapping defined on an open region d in e2 , with range r in e2 , and lets its jacobian be never zero in d. suppose further that c is a simple closed curve that , together with its interior ( recall the jordan curve theorem ) , lies in d , and that f is one - to - one on c. then the image t of c is a simple closed curve that , together with its interior , lies in r. furthermore , f is one - to - one on the closed region consisting of c and its interior , so that the inverse transformation can be defined on the closed region consisting of t and its interior . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ which i paraphrase as _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if the jacobian of a continuously differentiable two dimensional mapping is nonzero throughout an open region d and if the mapping is one to one on a simple closed curve c which lies in d , then the mapping is one to one on c and its interior . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ by using this theorem the following result can be established : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if two spectra with different maximum energies and such that suitably defined average energies of the spectra transmitted through an object are not equal over a region of the @xmath13 plane containing the points for the object , then the two measurements @xmath32 can be used to uniquely calculate the values of @xmath13 . _
_ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in order to establish this result , the jacobian must be shown to be nonzero throughout a region in the @xmath33 plane and the transformation must be shown to be invertible on a closed simple curve in this region . the first quadrant will be used as the region .
the closed curve consists of segments along the axes and a circle joining the ends of the segments , as shown in figure [ fig : closed - contour-4invertproof ] .
this is a region of theoretical and practical importance because a basis set consisting of the attenuation coefficients of the calibration materials is usually used , since only positive equivalent thicknesses of the calibration materials can be used , this region must contain all the measured values . . ] since @xmath32 is the logarithm of the measurements , the components of its jacobian are @xmath34de}{\int s_{i}(e)exp[-a_{1}f_{1}(e)-a_{2}f_{2}(e)]de}\quad i , j=1,2 \end{array}\label{eq : jij}\ ] ] note that by defining two normalized spectra @xmath35}{\int s_{i}(e)exp[-a_{1}f_{1}(e)-a_{2}f_{2}(e)]de}\quad i=1,2\ ] ] the jacobian matrix can be written as @xmath36\ ] ] where @xmath37 denotes the average using the normalized spectrum @xmath38 .
thus the condition for the determinant of the jacobian not being zero is equivalent to @xmath39 this condition must be tested for individual spectra . in order to complete the proof of invertibility
, the transformation must be shown to be invertible on a closed curve in the domain .
the simple cases discussed at the beginning of this section may be used for this proof .
the parts of the curve along the axes are special cases of the single material case .
each axis corresponds to different thicknesses of one of the basis materials if the attenuation coefficient of the calibration material is used as a basis function .
if other basis functions are used , the coordinates can be transformed to the attenuation coefficients of real materials and the proof applies in the transformed coordinates .
the circle of large radius is an approximation of the single energy case . for large radius
, there will be high attenuation . with beam hardening
, the transmitted spectrum with large attenuation therefore approaches the two monoenergetic spectra case where the energies are the maximum energies in the spectra .
if the maximum energies are different , we can approach the known invertible monoenergetic case arbitrarily closely by making the radius larger and larger .
in a previous paper@xcite , i used the central limit theorem of renewal processes@xcite to show that , for recording times much greater than the mean inter - event time , the expected value of recorded counts with pileup approaches @xmath40 in this equation , @xmath29 is the expected number of photons incident on the detector during the measurement time and @xmath41 is the expected number of photons arriving during the dead time @xmath42 .
if @xmath43 is the average rate of photon arrivals then @xmath44 . if @xmath45 is the measurement time then @xmath46 .
defining @xmath47 , @xmath48 and the expected recorded counts are @xmath49
in this section i give a proof that we can invert data consisting of measurements with two x - ray spectra with a photon counting detector with pileup .
an example would be making two sequential measurements of an object using an x - ray with switching the voltage .
note that this does not prove that any two spectrum measurement with pileup is invertible .
for example with two bin photon counting with pha , pileup causes the recorded spectrum to change so the assumptions of this section would not be met .
the proof is analogous to the proof for the measurements with no pileup described in the previous sec .
[ sec : invertibility-2dim-2spect - no - pileup ] .
i will show the transformation is invertible on the path in the first quadrant shown in fig .
[ fig : closed - contour-4invertproof ] .
the previous proof for invertibility on the circular segment is also applicable with pileup since , for large thicknesses , the count rate is very low so the pileup parameter @xmath41 is essentially equal to zero and the pileup counts are the same as those without pileup . the remaining part of the proof is to show that the data are invertible on the paths from the origin along the coordinates axes .
the equation for the expected value of the recorded number of photons with pileup is @xmath50 differentiating this equation along the axes @xmath51 where a prime denotes a derivative with respect to object thickness .
factoring the equation @xmath52 since the denominator is always positive and , from eq .
[ eq : lambda - prime ] , @xmath53 is always less than zero , the derivative of the recorded counts with respect to object thickness is never equal to zero so the transformation of the recorded counts with pileup is invertible along the coordinate axes .
this completes the proof if we can show that the jacobian of the transformation is never zero inside the contour .
this depends on the spectra but for commonly used spectra such as from x - ray tubes with different voltages , experience has shown that the transformation is invertible in the first quadrant .
the data acquisition model used is unrealistic most systems with photon counting detectors would also use pha .
nevertheless , it provides an example of invertibility with pileup . as discussed in my previous paper@xcite , due to the exponential probability distribution of x - ray photon inter - arrival times ,
pulse pileup is a fundamental effect
. it will always be present in photon counting systems no matter how small the response time .
pileup has accurate analytical models and its analysis may lead to an understanding of invertibility with the other detector defects such as charge trapping and sharing , polarization , and incomplete photon energy deposition due to compton scattering and k radiation escape@xcite .
a proof is given of the invertibility of two dimension basis set line integrals from two spectrum , photon counting data with pileup .
r. e. alvarez and a. macovski , `` energy - selective reconstructions in x - ray computerized tomography , '' _ phys .
* 21 * , 73344 , ( 1976 ) .
[ online ] .
available : http://dx.doi.org/10.1088/0031-9155/21/5/002 m. overdick , c. baumer , k. j. engel , j. fink , c. herrmann , h. kruger , m. simon , r. steadman , and g. zeitler , `` towards direct conversion detectors for medical imaging with x - rays , '' _ ieee trans .
sci . _ * nss08 * , 15271535 , ( 2008 ) .
[ online ] .
available : http://dx.doi.org/10.1109/nssmic.2008.4775117 r. e. alvarez , `` energy dependent information in x - ray imaging
part 1 the vector space description , '' _ stanford university information systems laboratory - unpublished _ , * * ( 1982 ) .
[ online ] .
available : http://dx.doi.org/10.13140/rg.2.2.19355.46887 k. taguchi , m. zhang , e. c. frey , x. wang , j. s. iwanczyk , e. nygard , n. e. hartsough , b. m. w. tsui , and w. c. barber , `` modeling the performance of a photon counting x - ray detector for ct : energy response and pulse pileup effects , '' _ med .
phys . _ * 38 * , 10891102 , ( 2011 ) .
[ online ] .
available : http://dx.doi.org/10.1118/1.3539602 r. e. alvarez , `` signal to noise ratio of energy selective x - ray photon counting systems with pileup , '' _ med .
_ * 41 * , no . 11 , 111909 , ( 2014 ) .
[ online ] .
available : http://dx.doi.org/10.1118/1.4898102 | in the alvarez - macovski@xcite method , the line integrals of the x - ray basis set coefficients are computed from measurements with multiple spectra .
an important question is whether the data are invertible .
this paper presents a proof that measurements with two spectra and a photon counting detector with pileup are invertible .
key words : photon counting , dead time , pileup , spectral x - ray , dual energy , energy selective = 1 |
the experimental program on deeply bound pionic states has revived the interest on the interaction of pions with nuclei .
recently the attention has focused on the charge exchange scattering length @xmath1 @xcite .
the fit to these data @xcite suggests a large enhancement of @xmath1 in nuclei , as anticipated in ref .
it is the purpose of this letter to investigate this question within a non - linear representation of the sigma model keeping the scalar degree of freedom .
the advantage of this model is that it allows an explicit evaluation of the contribution of the nuclear pion gas in a way which satisfies the constraints imposed by chiral invariance . at the same time it keeps the extra degree of freedom associated with the scalar meson so as not to exhaust the physics only with pions .
the quantity that we investigate is the charge exchange amplitude of a quasi - pion ( _ i.e. _ , on mass shell in the medium ) on a nucleon embedded in infinite nuclear matter .
we will derive the in - medium correction to first order in the nucleon density .
our emphasis is not so much on a comparison with data but rather on the derivation itself of the in - medium value for @xmath1 and on some important questions of principle arising in comparison with other theoretical works .
our starting point is based on the linear sigma model but reformulated as in ref .
@xcite . the original sigma field @xmath2 and
pion fields @xmath3 are eliminated in favor of a chiral invariant scalar field , @xmath4 , and a new pion field @xmath5 according to : @xmath6~.\ ] ] as proposed in ref .
@xcite , the fluctuating scalar field @xmath7 , which physically describes the fluctuations of the radius of the chiral circle around its vacuum value @xmath8 , can be identified with the sigma meson of relativistic theories of the walecka type .
the function @xmath9 selects the particular realization of the model . changing @xmath10 amounts to a redefinition of the pion field , which should not affect physics . in the following we keep @xmath11 arbitrary and check that the final results for physical observables do not depend on this parameter .
the lagrangian as given by eqs.(19 - 30 ) of ref .
@xcite writes ( notice that the notations @xmath12 for the chiral invariant scalar field are now replaced by @xmath13 ) : @xmath14 where we have defined ( with @xmath15 ) : @xmath16 this field transformation has the advantages of the non - linear realization but the radius of the chiral circle is not frozen and it keeps the degree of freedom associated with the scalar meson .
the new pion field couples derivatively ( to both the nucleon and the scalar field ) in such a way that the low - energy theorems are satisfied without need of cancellations . as for the symmetry breaking piece we have the following expression in terms of the new variables : @xmath17 in order to reproduce the experimental value of the isoscalar scattering length it is necessary , as in chiral perturbation theory , to introduce additional chiral invariant pieces contributing to order @xmath18 : @xmath19 where @xmath20 and @xmath21 is the four - velocity of the nucleon .
these @xmath22 and @xmath23 parameters have to be empirically determined .
finally the introduction of a form factor at the p - wave @xmath24 vertex allows the contact with the standard pion - nucleon and pion - nucleus phenomenology . in this model
the pion - nucleon sigma term , @xmath25 , receives two contributions , one from the scalar field of the nucleon and one from its pion cloud : @xmath26 the scalar contribution is given by : @xmath27 where @xmath28 is the scalar meson - nucleon coupling constant .
in the nuclear medium the relative amount of restoration is linked to the mean scalar field @xmath29 . to
leading order in density it is given by : @xmath30 the pion cloud contribution ( with its important @xmath31 piece ) has been explicitly calculated with a model form factor .
the resulting value is @xmath32 25 mev @xcite , about half of the total value . in order to link it to the chiral perturbation
result we expand it in orders of the pion mass : @xmath33 the coefficient @xmath34 depends on the model ( _ i.e. _ , on the form factor ) .
it is related to the coefficient @xmath35 of chiral perturbation theory by @xmath36 .
the leading non analytical contribution , @xmath37 , is instead given by chiral symmetry alone : @xmath38 its numerical value is @xmath39 mev , to be compared with the value of the full sigma term which is @xmath40 mev .
we stress that such a model pion cloud calculation has been successfully used to extrapolate lattice data into the low quark mass , _
i.e. _ , to the low pion mass sector @xcite . introducing the scalar density of nuclear pions , @xmath41 , the relative amount of restoration from the pion cloud
is given to leading order in density by : @xmath42 we now turn to the s - wave isoscalar pion self - energy for pions of zero three - momentum . to linear order in density
it is simply given by the vacuum @xmath0 @xmath43-matrix multiplied by the density .
the various contributions are depicted in fig .
[ piself ] . the pion loop contribution ( fig .
1a ) , including vertex corrections not shown in fig .
[ piself]a , is derived in refs .
it involves the pionic piece of the sigma term .
as for the scalar exchange contribution ( fig .
1b ) it is related to the scalar piece of the sigma term . fig .
1c is immediately obtained from the lagrangian given in eq .
( [ extra ] ) and fig .
1d is the born term contribution with pseudo - vector coupling .
the final result to leading order in density reads : @xmath44 note that the pion self - energy is not directly an observable and it depends on the representation through the @xmath45 factor : @xmath46
the energy dependence of the pion self - energy implies a wave - function renormalization of the in - medium pion , as noticed in several works @xcite .
the particular importance of the energy dependence for the interpretation of the deeply bound pionic states has recently been emphasized in ref .
the pion propagator has the form @xmath47 . from the expression for the self - energy , eq .
( [ pimodel ] ) , we obtain for the residue : @xmath48 the value of the isoscalar scattering length , @xmath49 , provides a relation between the different parameters entering this expression . for the nucleon it is given by : @xmath50 according to recent accurate data @xcite , this value is compatible with zero .
this near - vanishing of @xmath49 ( believed to be fortuitous ) translates into the following expression for the residue : @xmath51 the residue provides a first source for the renormalization of the isovector scattering length @xmath1 . notice that if the pion loop correction ( the last two terms in eq .
( [ resmodel ] ) ) is ignored one recovers the original proposal of ref .
. however this description would be incomplete since , as previously discussed , the pion cloud piece of the sigma term is certainly not negligible .
in addition a description of the in - medium modification with the only influence of the residue is clearly unsatisfactory since the result then depends on the representation .
indeed there is another source of renormalization , which to our knowledge has previously been ignored .
it arises from the pion loop correction of the isovector pion - nucleon amplitude , inherent to the non - linear realization . the isovector piece ( weinberg - tomozawa ) of the chiral lagrangian is given by : @xmath52 where the second expression arises from an expansion to fourth order in the pion field .
after the appropriate isospin averaging , one obtains an effective weinberg - tomozawa lagrangian which is modified in the medium by one - pion loop correction ( see fig .
[ piwt ] ) : @xmath53 the isospin - antisymmetric amplitude is proportional to the pion energy @xmath54 . in the vacuum @xmath55 , and @xmath56 . for the quasi - particle in the medium
the energy is the pion effective mass @xmath57 .
since we take @xmath49 to be zero , to first order in the density the effective and bare pion masses are equal .
according to the previous discussion the in - medium renormalization of @xmath1 is then : @xmath58 which gives : @xmath59 notice that once the two influences are taken into account the representation dependence has disappeared , an important consistency check of our result .
coming to a numerical evaluation , the parameters of the model , in particular the sigma mass , can be fixed by a simultaneous fit to the saturation curve and to pion - pion phase shifts @xcite .
a typical value is @xmath60 mev , yielding @xmath61 mev .
taking also @xmath62 mev , we have : @xmath63 this leads to the estimate : @xmath64 even if the coefficient @xmath65 may be slightly changed if the relative weight between the pion cloud and the scalar contribution to the sigma term is changed , the enhancement is more moderate than the original proposal @xcite which ignores the pion loop correction ( in that case the coefficient is of the order of @xmath66 depending on the precise value of the sigma term ) . in our case
this pion loop correction partly cancels the effect of the pion - nucleon sigma term .
the pion - self - energy has been calculated using a chiral perturbation approach . to leading order in density
, it reads @xcite : @xmath67 one can make a one - to - one correspondence between this chipt result and the result obtained in our model ( eq . [ pimodel ] ) .
the first term ( @xmath25 ) and the third term ( born term ) are identical in both expressions of the self - energy .
the fourth terms in both expressions are also equivalent since the comparison between the two just defines the quantity @xmath68 in the particular model that we use . as for the representation
dependent coefficient @xmath69 which appears in the last term of eq .
[ pichipt ] , it has to be be identified with @xmath70 .
however there is an essential difference both at the conceptual and quantitative levels .
the second term and the fifth one ( representation dependent term ) involve the lnac piece of the pion loop , at variance with the model calculation where the full pion loop appears .
we will return later to this difference and we study for the moment the consequences for the @xmath1 parameter . following the same steps as in section iii , with the constraint that the isoscalar scattering length vanishes , we obtain for the residue : @xmath71 taking , as in ref .
@xcite , @xmath72 ( _ i.e. _ @xmath73 ) , the influence of the residue on the @xmath1 parameter becomes : @xmath74 which yields a larger enhancement than our model calculation .
however this result is not representation independent .
the pion - loop correction in the weinberg - tomozawa amplitude which is not taken into account in the chipt calculation should be added . if this is done in a consistent way one should retain only the lnac piece of the pion loop in order to reach the representation independent following result : @xmath75 this is numerically very close to what is obtained from the prescription used in ref .
@xcite but still difficult to reconcile with our model result .
the crucial difference is the fact that , in our model , the pion loop always appears as a whole and the lnac can never appear as an isolated piece . in other words
the model involves the combination @xmath76 rather than @xmath77 alone .
the two quantities are of comparable magnitude but of opposite signs .
the difference between the two approaches is associated with different off - mass shell behavior of the @xmath0 amplitude . as a further test of our method , in the pcac representation in which the relation @xmath78 holds
, the soft pion amplitude ( obtained by taking @xmath79 ) should be proportional to @xmath25 . in order to obtain the pcac representation it is sufficient to send
the sigma mass to infinity ( the sigma term then has only the pionic contribution ) and to take @xmath80 ( _ i.e. _ , @xmath81 ) . in that case one
recovers at the soft pion point the venerable low - energy theorem since : @xmath82 instead with the prescription of eq .
( [ pichipt ] ) for the chipt off - shell amplitude , one gets : @xmath83 the two terms on the r.h.s . being negative can not cooperate to reproduce the positive sigma term , in conflict with the soft pion theorem .
we have studied the in - medium renormalization of the isovector amplitude in the linear sigma model .
we use it in a non - linearized representation in which the radius of the chiral circle is not frozen , which preserves the sigma degree of freedom . in this approach
the nucleon sigma commutator is built of a scalar meson exchange with the condensate and a two - pion exchange one , the pion cloud contribution .
we find , as in our previous works on other in - medium quantities , that the independence on the representation is achieved through a combination of two effects : i ) the residue of the pion propagator arising from the energy dependence of the pion self - energy , ii ) the influence of the pion loops on the weinberg - tomozawa amplitude ( which must also apply to the chiral perturbation approach of refs .
@xcite so as to make it independent on the representation ) .
our description results in a more moderate renormalization of @xmath1 : at normal nuclear density , an enhancement by 18% , which is less than the prediction of ref .
@xcite or the one which can be derived from the results of ref.@xcite .
the difference between our model and the chipt approaches lies in the role of the pion loops .
our treatment for these is identical to the one employed in ref .
@xcite for the pion gas in a heat bath .
the results of ref .
@xcite , when taken in the chiral limit , reproduce the chiral perturbation expansion up to second order .
the only difference between the present calculation and the description of the pion gas in a heat bath amounts to the ( physically natural ) replacement of the scalar density of thermally excited pions by that of the nuclear pions , a positive quantity . on the contrary , in the chiral approach of refs .
@xcite only the leading non - analytical term appears in the self - energy ( beyond what is implicitly contained in the nucleon sigma commutator ) . since @xmath77 is negative and comparatively large , the numerical consequences which follow explain the difference in the renormalization factor . it will be interesting to exactly clarify why , in the chiral approach , @xmath77 , which is in fact a piece of the pion density , dissociates from the rest , while the whole pion density ( with its simple physical interpretation ) naturally enters in our approach .
the nucleon size , which is crucial in any model evaluation of the nuclear pion loop , introduces a scale which does not seem to be fully incorporated in this chipt approach .
the model calculations of the pion loop which explicitly takes into account the nucleon size ( _ i.e. _ , in practice the form factor ) have proven their relevance and usefulness .
for instance they give an excellent fit to the quark mass dependence of the lattice results allowing a convincing extrapolation to the physical mass region @xcite .
the actual fits of recent data on deeply bound pionic states suggest a larger enhancement of @xmath1 than our prediction @xcite . in this context
we want to point out that the comparison between our result and the the fit to pionic atoms data is not direct .
first of all we consider only strong interactions and ignore effects due to coulomb forces , while these are crucial to bind the pion within the pionic atoms on which the fit is based . they are discussed in ref .
moreover not all renormalization effects are included in our approach .
for instance , we did not consider pion rescattering on two correlated nucleons .
it is known to provide a sizeable part of the repulsive isospin - symmetric potential . since it is not explicitly introduced in the fits to data for the charge exchange potential
, it has to be included as a medium renormalization of @xmath1 .
this effect can be evaluated from the formulae given in ref .
@xcite : @xmath84 where @xmath85 is the fermi momentum of the nucleons . at normal density , it amounts to a decrease of @xmath1 by @xmath86 , _
i.e. _ , it does not help to explain a larger enhancement .
it would be interesting to attempt to fit the pionic atom data with a more moderate enhancement of the @xmath1 parameter .
if the larger enhancement is confirmed it would be an incentive to push the investigations further .
one might consider the influence of higher order effects in the density , which could upset the nearly total cancellations present in the isospin symmetric scattering length .
for instance , the pionic piece of the sigma commutator should be replaced by the fully dressed one in which the pion line is dressed by nucleon - hole and @xmath31-hole excitations screened by short range correlations . at normal nuclear density
the effect leads to a moderate enhancement ( of the order of 25% ) of @xmath41 @xcite .
the many - body effects in the other contributions should be similarly investigated , including those concerning the phenomenological parameters @xmath22 and @xmath23 , which would require a model for these quantities .
the question of the in - medium modification of the charge exchange @xmath0 amplitude first raised by weise is indeed quite challenging on the theoretical side as well as on the experimental one .
it is likely that more physics remains to be understood in this problem .
99 h. geissel et al . , phys . rev .
88 ( 2002 ) 122301 ; nucl .
a 663 ( 2000 ) 206 .
p. kienle and t. yamazaki , phys .
b 514 ( 2001 ) 1 .
w. weise , nucl .
a690 ( 2001 ) 98c . n. kaiser and w. weise , phys .
lett . b512 ( 2001 ) 283 .
e. friedman , phys .
b524 ( 2002 ) 87 ; nucl .
phys a710 ( 2002 ) 117 .
kolomeitsev , n. kaiser , and w. weise , nucl - th/0207090 .
t. yamazaki , plenary talk given at panic02 , osaka ( japan ) 28 sept- 4 oct 2002 ; to be published in nucl .
a. g. chanfray , m. ericson , and p.a.m .
guichon , phys .
rev c63 055202 .
i. jameson , a.w .
thomas , and g. chanfray , j. physics g18 ( 1992 ) l159 .
m. birse , j. mc govern , phys .
b292 ( 1992 ) 242 d.b .
leinweber , a.w .
thomas , k. tsushima , and s.v .
wright , phys .
d61 ( 2000 ) 074502 .
g. chanfray , m.ericson , and j. wambach phys .
b388 ( 1996 ) 673 . g. chanfray and d. davesne , nucl .
a646 ( 1999 ) 125 .
g. chanfray , m.ericson , and m. kirchbach , modern phys .
a 9(1994 ) 279 . v. thorsson and a. wirzba , nucl .
a589 ( 1995 ) 633 .
g. chanfray , j. delorme , and m. ericson , nucl .
a637 ( 1998 ) 421 .
a. wirzba , j. oller , and u.g .
meissner , ann .
( 2002 ) 27 h.c .
schrder et al . , eur .
( 2001 ) 473 .
g. chanfray , plenary talk given at panic02 , osaka ( japan ) 28 sept- 4 oct 2002 ; to be published in nucl .
a. g. chanfray , m. ericson , and m. oertel , work in preparation .
t.s park , h. jung , and d.p .
min , nucl - th/0101064 .
m. ericson and m. rho , phys . rep .
5c ( 1972 ) 59 .
g. chanfray , d. davesne , j. delorme , m. ericson , and j. marteau , eur .
j. a8 ( 2000 ) 283 . | we study the in - medium modification of the isovector @xmath0 amplitude using a non - linear representation of the sigma model but keeping the scalar degree of freedom .
we check that our result does not depend on the representation .
we discuss the connection with other approaches based on chiral perturbation theory . |
initially through trace and eit observations , and now ongoing with aia imagery ( e.g. , * ? ? ?
* ) , the loops of magnetic field that often overlie magnetic active regions are seen to oscillate with frequencies of typically 24 mhz ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the attribution of these oscillations to the modes of a one - dimensional cavity , and the observational identification of overtone frequencies of this cavity , has given birth to the field of coronal seismology ( see nakariakov & verwichte 2005 and the references therein ) . the promise offered by observable resonant oscillation frequencies has raised hopes for determining physical parameters of coronal loops which cannnot otherwise be measured directly .
although multiple claims have been made for the observation of sausage waves in post - flare coronal loops ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , the majority of the observed oscillations useful for seismological purposes are kink oscillations @xcite .
the dispersion in the observed overtones , compared to theoretical predictions , have been used mainly to estimate the mean magnetic field strength ( e.g. , * ? ? ?
* ) , it s longitudinal gradient @xcite , or the density stratification inside and outside the loop @xcite .
observational efforts are now focused on improving the measurement of oscillation eigenfrequencies by reducing the observational errors and by analyzing additional loops .
a relevant overview of observational measurements can be found in @xcite . on the theoretical side ,
advancements are steadily being made by including additional physical effects with the aim of developing a general model of coronal loops and their oscillations .
most of this work is based on the pioneering calculations of @xcite , which were presciently performed prior to the first observations of coronal loop waves .
this work revealed the mhd wave modes of a translationally invariant magnetic cylinder .
since then @xcite and @xcite have investigated modifications to the eigenfrequencies that arise from density stratification along the loop .
the expected dispersion has been compared to measured values with the goal of constraining coronal loop models ( see also * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
further , the quantitative effects of loop geometry were investigated by @xcite and @xcite , with particular focus on the curvature of the loop axis .
for example , @xcite have shown that the frequency shift induced by curvature appears at second order in the ratio of tube radius to the loop length , which is believed to be quite small for the majority of loops .
they and others have also suggested that since the eigenfrequencies for both horizontal and vertical @xcite polarizations are identical , there is no distinction between the oscillations with different polarization ( e.g. , * ? ? ?
* ) , unless of course the tube is not locally axisymmetric @xcite . here
, our goal is to self - consistently include the effects of gravity and curvature , both in the establishment of the static equilibrium of a coronal loop and in the propagation of kink waves along the loop .
we accomplish both of these goals by exploiting the thin - tube approximation , whereby lateral variation across the tube is ignored . under this assumption
the equilibrium shape of the coronal loop is established by a balance between the forces of magnetic buoyancy and magnetic tension .
the eigenmode problem is reduced to a one - dimensional wave equation which we solve through perturbation theory .
the paper is organized as follows : in section 2 we present the equation of motion appropriate for a thin magnetic fibril . in section 3
we derive the equilibrium height and shape of the loop , while section 4 details the solution to the kink wave eigenmode problem . in section 5 ,
we discuss the significance of our findings and the implications of our base assumptions .
finally , we state our findings and conclusions in section 6 .
we will treat a coronal loop as a thin magnetic fibril embedded within a field - free coronal atmosphere .
we recognize that both of these assumptions are suspect . even though the thin tube approximation may only apply to a subset of coronal loops , it provides mathematical tractability .
the assumption that the loop is an isolated magnetic structure is probably never strictly valid , as the entirety of the solar corona is magnetically dominated and most loops are an integral part of larger magnetic structures . in fact , most loops are probably just selectively illuminated field lines amongst many that form magnetic arcades . despite these objections
, we adopt an isolated fibril model for convenience .
the consequences of these assumptions are discussed in more detail in section [ sec : discussion ] .
the forces acting on an isolated , thin magnetic fibril have been previously derived by @xcite through averaging the mhd momentum equation over the cross - sectional area of the tube . in terms of the acceleration of the tube , the resulting equation of motion
is given by @xmath3 { { { \mbox{{\boldmath$\hat{s}$ } } } } } + c^2 { { \mbox{{\boldmath$k$ } } } } + d \left({{{\mbox{{\boldmath$\hat{s}$ } } } } } \times { { \mbox{{\boldmath$g$}}}}\right ) \times { { { \mbox{{\boldmath$\hat{s}$ } } } } } \ ; , \ ] ] where all fluid properties represent cross - sectional averages and @xmath4 and @xmath5 are the instantaneous tangent unit vector and curvature vector of the tube s axis .
the gravitational acceleration is @xmath6 and @xmath7 is the gas pressure .
the longitudinal forces ( those in the @xmath4 direction ) are just the pressure and gravitational forces that act parallel to the magnetic field .
the last two terms on the right - hand side are the transverse forces of magnetic tension and buoyancy with @xmath8 being the kink wave speed and @xmath9 the tube s fractional overdensity compared to its surroundings , @xmath10 the densities @xmath11 and @xmath12 are , respectively , the mass density internal and external to the tube and @xmath13 is the axial magnetic field strength of the fibril .
the transverse forces ( tension and buoyancy ) in equation ( [ eqn : spruit ] ) include the effects of enhanced inertia , " which represents the backreaction caused by the external medium when transverse motions of the tube move surrounding fluid out of the way .
the form of the enhanced inertia is the same as propounded by @xcite , whereby the density in the inertial term is simply augmented , @xmath14 .
this form has inspired much controversy ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , but faithfully reproduces the proper kink wave speed and is consistent with all alternate formulations as long as ( 1 ) the background atmosphere lacks rotational shear and magnetic field , ( 2 ) flow along the tube is not permitted , and ( 3 ) correction terms that are quadratic in the fluid velocity are ignored .
we neglect the spherical geometry of the solar atmosphere and assume that the corona can be treated as plane - parallel with constant gravity @xmath6 .
we employ a cartesian coordinate system , with the @xmath15@xmath16 plane corresponding to the photosphere and the @xmath17 coordinate increasing upwards ( i.e. , @xmath18 ) .
the equilibrium position of the coronal loop is confined to the @xmath15@xmath17 plane with the endpoints of the loop anchored in the photosphere at the coordinates ( @xmath19,0,0 ) .
the height of the loop above the photosphere is given by the function @xmath20 , such that the loop s axis is traced by the position vector @xmath21 .
figure [ fig : geometry ] provides an illustration of the geometry of the loop and the coordinate system .
in addition to this cartesian coordinate system , we define a local set of frenet coordinates for the equilibrium position of the loop .
let @xmath22 denote the arclength along the loop measured from the footpoint located at @xmath23 .
the longitudinal unit vector , or the tangent vector that lies parallel to the loop s axis pointing in the direction of increasing @xmath22 , is denoted @xmath24 . within the @xmath15@xmath17 plane and everywhere perpendicular to @xmath24 is the principle normal @xmath25 .
we shall soon see that for the loop model we adopt , only convex loops are stable .
therefore , the curvature vector @xmath26 always points inward and downward in the direction of the principle normal @xmath25 , with a magnitude equal to the reciprocal of the local radius of curvature @xmath27 of the loop s axis .
the third orthogonal direction is given by the binormal unit vector @xmath28 , which is everywhere constant @xmath29 . in what follows
we will need the standard geometrical equations which describe the frenet unit vectors , @xmath24 and @xmath25 , and the radius of curvature @xmath30 in terms of the loop height , @xmath31 ^ 3}{z_0^{\prime\prime}(x ) } \ ; , \label{eqn : rcurv } \\ \nonumber \\
s^\prime(x ) & = & \sqrt{1 + \left [ z_0^{\prime}(x ) \right]^2 } \ ; , \label{eqn : sprime}\end{aligned}\ ] ] where primes denote differentiation with respect to the photospheric coordinate @xmath15 . since a loop in a state of static equilibrium
is confined to the @xmath15@xmath17 plane , none of the forces on the fibril have a component in the binormal @xmath16-direction and we only need to consider two components of the equation of motion ( [ eqn : spruit ] ) , @xmath32 in these two equations the subscript 0 indicates a background , equilibrium quantity within the loop . the triple cross - product appearing in the buoyancy term in equation ( [ eqn : spruit ] ) has been reduced to a simpler form by making use of the identities @xmath33 .
equation ( [ eqn : equil_s ] ) expresses the balance of forces in the tangential or axial direction @xmath24 , whereas equation ( [ eqn : equil_k ] ) describes the balance in the transverse direction of the principle normal @xmath25 .
the axial equation is simply hydrostatic balance along magnetic field lines which we will satisfy trivially by assuming that the loop and the surrounding corona are both isothermal .
furthermore , since the tube is thin and thermal diffusion wipes out lateral variations , the temperatures inside and outside the tube are identical , @xmath34 .
all densities and pressures ( both gas and magnetic ) therefore vary exponentially with height @xmath17 with a common scale height @xmath35 .
a consequence of this thermal uniformity is that the kink speed @xmath36 , the plasma - parameter @xmath37 , and the overdensity @xmath38 are all constants .
the shape of the loop @xmath20 is constrained by the balance of the transverse forces of magnetic tension and magnetic buoyancy , which is quantified by equation ( [ eqn : equil_k ] ) . since the temperatures inside and outside the loop are identical , and the external fluid is field free , the tube must be partially evacuated ( @xmath39 ) in order to ensure pressure continuity across the lateral boundary of the flux tube .
thus , magnetic buoyancy tries to lift the loop higher into the atmosphere and magnetic tension arising from the convex curvature of the loop tries to hold it down .
since the kink speed @xmath36 is constant for our isothermal model , all variation along the loop in the magnitude of the magnetic tension force comes from changes in the radius of curvature @xmath30 .
similarly , the buoyancy only varies because the transverse direction @xmath25 rotates relative to the vertical @xmath40 .
if we insert equations ( [ eqn : k0 ] ) and ( [ eqn : rcurv ] ) into equation ( [ eqn : equil_k ] ) , we obtain a nonlinear ode for the height of the loop @xmath20 , @xmath41 where we have defined a magnetic bond number , @xmath42 the magnetic bond number can be written in a more recognizable form if we multiply the numerator and denominator by the enhanced inertia @xmath43 , @xmath44 in this expression the surface tension that normally appears in the denominator of the bond number has been replaced by magnetic tension .
the magnetic bond number @xmath0 is a signed , dimensionless number that embodies the relative importance of the buoyancy and magnetic forces .
the magnetic bond number appears in calculations of the rayleigh - taylor instability when one of the fluid layers is filled with a horizontal field , providing the critical wavenumber below which instability ensues , @xmath45 @xcite . for our isothermal model ,
@xmath0 is a constant and furthermore it can be expressed as a ratio of length scales , in particular the ratio of the footpoint separation to the pressure scale height @xmath46 .
this property is easily derived by realizing that the combination of the perfect gas law and pressure continuity across the lateral surface of the tube allows one to express the magnetic pressure inside the tube as a ratio of the density difference to the fluid temperature .
when this expression is inserted into equation ( [ eqn : epsilon_variant ] ) , the density difference cancels in the numerator and denominator , leaving just the pressure scale height in the denominator ( which is proportional to the temperature ) .
since @xmath0 is a constant , equation ( [ eqn : eqode ] ) can be integrated directly to obtain the derivative of the height function @xmath47 .
similarly , this expression for the derivative can be subsequently integrated to obtain the height function itself @xmath20 .
we choose the constants of integration such that @xmath48 and @xmath49 .
a summary of the resulting properties of the loop is provided below :
@xmath50 the last of these equations was obtained by integrating equation ( [ eqn : sprime ] ) , with the constant of integration chosen such that @xmath51 .
figure [ fig : loopmodel ] displays solutions for several different values of @xmath0 . for reference
, the dashed curve corresponds to a semi - circle , clearly illustrating the variable radius of curvature for the loops .
the length of the loop @xmath52 is obtained by inserting the footpoint position @xmath53 into equation ( [ eqn : pathlength ] ) , @xmath54 two interesting limits of this equation exist . as @xmath55 the length of the loop converges to the footpoint separation @xmath56 .
this arises because the loop becomes straight , flat , and confined to the photosphere . as @xmath57 the loop length diverges logarithmically because the height of the loop grows without bound
this can be seen by recognizing that the loop reaches its apex at its center ( @xmath58 ) , therefore achieving a maximum height @xmath59 one can easily see that @xmath60 corresponds to a logarithmic singularity in the height . clearly the height @xmath20 should be a positive function for the range @xmath61 . with a little thought , from equation ( [ eqn : z0 ] ) we can see that this is only possible if @xmath62 .
therefore , for constant @xmath36 only underdense tubes ( @xmath39 ) form stable loops and those loops are comprised of a single arch without inflection points ( @xmath63 ) .
furthermore , for @xmath64 to remain positive , @xmath0 must be bounded and an equilibrium is only possible if the magnetic tension exceeds a threshhold , @xmath65 this condition for stability can also be used to constrain the footpoint separation . since , @xmath46 , stability requires that the footpoint separation be less than a critical value that depends on the corona s pressure scale height , @xmath66 typically , we do not need to be overly concerned about the limit of large buoyancy forces . for most coronal loops
the magnetic bond number @xmath0 is quite small . using typical values of the footpoint separation @xmath67 mm ,
kink wave speed @xmath68 mm s@xmath69 @xcite , overdensity @xmath70 ( appropriate for very small plasma parameter @xmath71 ) , and gravity @xmath72 mm s@xmath73 , we find from equation ( [ eqn : epsilon ] ) that @xmath74 . for small values of @xmath0 ,
the loop is short and low - lying with @xmath75 .
this geometry is realized because the field lines exit the photosphere at an oblique angle i.e .
, @xmath76 . as
@xmath0 becomes small , the angle between the fibril s axis and the photospheric surface approaches zero ( see figure [ fig : loopmodel ] ) .
kink waves cause the axis of the fibril to undulate ; therefore , we need to consider wave - induced modulation of the local direction vectors , @xmath4 and @xmath77 .
let the instantaneous position vector of the loop be described as a time - varying perturbation about a static equilibrium .
we will consider two transverse displacements of the axis , one displacement @xmath78 polarized in the direction of the principle normal @xmath25 ( which is confined to the @xmath15@xmath17 plane ) and a perpendicular displacement @xmath79 polarized in the binormal @xmath80 direction ( i.e. , horizontal and parallel to the photospheric plane ) , @xmath81 in the equation above , as before , the coordinate @xmath22 is the pathlength along the unperturbed position of the tube . if we carefully consider the perturbations to the unit vectors , we find to first order in the displacements @xmath82 since , the equilibrium configuration of the loop lacks curvature in the @xmath80 direction , the radius of curvature does not depend on the @xmath83 displacement , at least to linear order .
because of this , the two polarizations decouple , and @xmath84 and @xmath83 satisfy independent equations .
these two wave equations are obtained by noting that transverse oscillations of the thin fibril are largely incompressive ; therefore , the longitudinal component of the equation of motion , equation ( [ eqn : spruit ] ) , can be ignored .
differential equations describing the two polarizations of oscillation can be derived by inserting equations ( [ eqn : sdir])([eqn : rcurvwave ] ) into equation ( [ eqn : spruit ] ) , fourier transforming in time , and linearizing in the displacements , @xmath85 the influence of the geometry of the equilibrium loop model is felt only through the radius of curvature @xmath30 and through the parallel component of gravity @xmath86 .
the terms proportional to @xmath87 represent magnetic tension , with the radius of curvature corresponding to the contribution from the curvature of the equilibrium field lines .
the terms with @xmath88 are buoyancy forces and the inertial terms are those that depend on the temporal frequency @xmath89 .
these are quite general equations that describe oscillations for any loop equilibrium not just the isothermal model we consider here .
in fact , except for the inclusion of curvature , these are the same equations used by @xcite to study the effects of buoyancy and wave speed variations on mode frequencies .
let s first examine the tension terms in detail .
when these two equations are compared , we see that the curvature of the tube only appears in the equation for the normal oscillations .
this additional tension term from curvature typically has the opposite sign from the tension arising from undulations .
thus , the two tension forces that appear in the equation for normal oscillations act in opposition and we should expect that curvature of the loop reduces the eigenfrequencies .
further , if we consider the limit of small magnetic bond number @xmath0 , we can see from equation ( [ eqn : r0 ] ) that the radius of curvature is large @xmath90 .
therefore , the curvature of the tube should generate a second - order frequency shift @xmath91 . if we insert @xmath92 into equation ( [ eqn : zetaeqn_raw ] ) , and include the magnetic bond number dependence directly , we get @xmath93 a cursory examination of this equation would seem to indicate that for small @xmath0 , the frequency shift from buoyancy forces should be a first - order quantity . however , this is incorrect .
we have previously demonstrated that the equilibrium height and shape of the loop sensitively depend on the magnetic bond number , and in particular as @xmath55 the loop becomes flat and horizontal .
in fact , by using equations ( [ eqn : s0hat ] ) , ( [ eqn : sprime ] ) , and ( [ eqn : dz0 ] ) one finds @xmath94 since @xmath95 for small @xmath0 , it is now clear that buoyancy should cause frequency shifts that are second order in the magnetic bond number .
this makes sense , as curvature also enters at second order and buoyancy and curvature are balanced in the equilibrium configuration of the loop .
the quantities @xmath88 and @xmath30 depend on the equilibrium shape of the loop , which depends explicitly on the horizontal coordinate @xmath15see equations ( [ eqn : z0])([eqn : pathlength ] ) .
therefore , it is quite natural to make a change of variable from the arclength variable @xmath22 to the horizontal @xmath15 coordinate . after inserting the properties of the equilibrium , we find that the resulting odes lack first - derivative terms , @xmath96 \xi & = & 0 \ ; , \\ \nonumber \\ { \ensuremath { \frac{d^{2}{\zeta}}{d{x}^{2 } } } } + \frac{\omega^2}{c_0 ^ 2 } \ \frac{1}{\cos^2\left(\epsilon x / x\right ) } \ \zeta & = & 0 \ ; . \label{eqn : zetaeqn}\end{aligned}\ ] ] the absence of the first derivatives occurs because the tension and buoyancy forces in the equilibrium are in exact balance .
both of these equations have solutions in the form of associated legendre functions , @xmath97 \ ; , \\ \nonumber \\
\zeta(x ) = \left(\cos\theta\right)^{1/2 } \left [ a_3 p_{-1/2}^\mu\left(\sin\theta\right ) + a_4
q_{-1/2}^\mu\left(\sin\theta\right ) \right ] \ ; , \end{aligned}\ ] ] with @xmath98 the dependence on the frequency @xmath89 is buried in the upper index @xmath99 of the associated legendre functions .
therefore , these solutions are not particularly friendly , and applying boundary conditions to quantize the eigenfrequencies would require a nontrivial numerical computation . we avoid this difficulty by realizing that the magnetic bond number @xmath0 is a small dimensionless parameter , thus allowing the use of perturbation theory . the equations that describe oscillations in both the normal and binormal directions have very similar form .
in fact , a brief inspection of equations ( [ eqn : xieqn ] ) and ( [ eqn : zetaeqn ] ) reveals that the eigenfunctions of the two polarizations should be identical up though second order in @xmath0 .
this means that the derivation of the solutions for the two polarizations are similar ; therefore , we will demonstrate the perturbation analysis for only the normal component and simply state the result for the binormal component .
assume that the eigenfunction and eigenfrequency can be written as perturbation expansions , @xmath100 insert these expansions into equation ( [ eqn : xieqn ] ) and solve in the usual fashion for the zeroth - order eigenfunction and eigenfrequency , noting that the cosine appearing in the second term can be expanded for small argument , @xmath101 \ ; , \\
\nonumber \\
\omega_n & = & \kappa_n c_0 = \frac{n\pi c_0}{2x } \ ; .
omega_n}\end{aligned}\ ] ] the boundary condition of vanishing displacement at the footpoints @xmath102 has been applied .
if we now collect terms of the next order , we find the following inhomogeneous equation for the perturbed eigenfunction , @xmath103 the particular solution can be found by judicious guessing and by enforcing that the perturbed eigenfunction has the same symmetry about the center of the loop as the unperturbed eigenfunction , @xmath104 \xi_n^\prime(x ) \ ; .\ ] ] the perturbed eigefrequency is fixed by reinforcing that the eigenfunction vanishes at the footpoints for all orders of expansion @xmath105 , @xmath106 if we now put all the pieces together , to second order in the small parameter @xmath0 , the eigensolution for the normal displacement is given by @xmath107 + o(\epsilon^4 ) \ ; , \\ \nonumber \\
\omega_n^2 & = & \kappa_n^2 c_0 ^ 2 \left[1 - \epsilon^2 \left ( \frac{1}{3 } + \frac{1}{2\kappa_n^2 x^2 } \right ) + o(\epsilon^4)\right ] \ ; . \label{eqn : eigfreq_norm}\end{aligned}\ ] ] by the same procedure the eigensolution for the binormal polarization can be obtained , @xmath108 \ ; . \label{eqn : eigfreq_binorm}\end{aligned}\ ] ] figure [ fig : eigfreq ] illustrates the resulting eigenfrequencies as a function of @xmath0 for displacements in both the normal and binormal directions . figure [ fig : eigfunc ] shows sample eigenfunctions .
in the previous sections we have modeled a coronal loop as an isolated , thin magnetic flux tube .
the effects of gravity and curvature have been self - consistently included , leading to a static equilibrium determined by the balance of magnetic buoyancy and tension .
we have computed the eigensolutions for transverse oscillations of both polarizations . in the following subsections
we will discuss in detail the effects that gravity and curvature have on the structure of the eigenfrequencies , assess the differences between the two polarizations of wave motion , and explore the implications of relaxing several of our operating assumptions .
there are only two free parameters in our model , the footpoint separation @xmath109 and the magnetic bond number @xmath0 .
the footpoint separation just provides a physical scale for the problem and the structure of the solutions do not depend upon it .
the solutions do depend intrinsically on the magnetic bond number , because the equilibrium shape of the loop is sensitive to the strength of buoyancy and magnetic tension . as the magnitude of the parameter @xmath0 becomes larger ( more negative ) , buoyancy forces increase in strength , which causes the loop to rise and
a corresponding increase in the curvature of the loop is needed to allow magnetic tension to bring the loop back into equilibrium .
this results in three ways by which the frequencies are modified : the loop lengthens , the restoring force of buoyancy increases , and the increased curvature modifies magnetic tension .
we can identify the first of these effects on the eigenfrequencies by expanding the loop length @xmath52 , equation ( [ eqn : length ] ) , for small @xmath0 , @xmath110 \ ; .\ ] ] if the loop were straight and without gravitational effects , the eigenfrequencies would be @xmath111 \ ; .\ ] ] as expected , an increase in the loop length ( increasing @xmath112 ) decreases the eigenfrequencies and the resulting fractional frequency shift is dispersionless and independent of mode order , @xmath113 this effect appears directly , in equations ( [ eqn : eigfreq_norm ] ) and ( [ eqn : eigfreq_binorm ] ) as the first term in the parentheses on the right - hand sides .
buoyancy appears in identical form in the equations describing each polarization of transverse oscillation ; both equation ( [ eqn : xieqn_raw ] ) and ( [ eqn : zetaeqn_raw ] ) have the same first - derivative term .
this common form , leads to identical perturbations to the eigenfrequencies due to the inclusion of buoyancy . from equations ( [ eqn : eigfreq_norm ] ) and ( [ eqn : eigfreq_binorm ] ) we can directly see that buoyancy increases the eigenfrequencies by the fractional amount , @xmath114 this might be a somewhat surprising result . a simple dimensional analysis of the equation of motion would indicate that buoyancy causes frequency shifts that are first order in the magnetic bond number @xmath0 , and indeed this is what was found in @xcite . however , the buoyancy force in the wave equation is proportional to @xmath0 times the geometrical factor @xmath115 .
this geometrical factor also depends on @xmath0 because the shape of the coronal loop is sensitive to @xmath0 .
low magnetic bond numbers correspond to loops that are low - lying and nearly horizontal , and because of this the geometrical factor is first order in the magnetic bond number , and the buoyancy force in total is second order .
the last effect , curvature of the loop , modifies the magnetic tension than can act as a restoring force .
it only appears in the equation for the normal polarization , and its sign indicates that it counteracts the local curvature induced by the wave motions ( i.e. , the second - derivative of the displacement ) .
therefore , we would expect curvature of the background loop model to decrease the eigenfrequencies .
this does indeed occur , as we can see directly from the difference between the frequencies of the normal and binormal polarizations .
the curvature of the loop results in a negative frequency shift that is twice as large in magnitude as the effect of buoyancy , @xmath116 the frequency shifts induced by both buoyancy and curvature enter at second order in the magnetic bond number .
that both enter at the same order is a simple consequence of the balance between buoyancy and tension in the equilibrium .
that they both enter at second order could easily be predicted . from dimensional arguments alone ,
the shift in the square frequency due to curvature must scale with the radius of curvature as @xmath117 .
since the radius of curvature is inversely proportional to @xmath0 , the frequency shift must be second order in @xmath0 .
a consistent result for the curvature was found by @xcite where they solved for fast waves in a toroidal coordinate system in the absence of gravity .
the equilibrium field was assumed to be purely toroidal and a coronal loop modeled by a partial section of the torus . with an extensive mathematical analysis they concluded that the curvature of their torus induces a frequency shift that vanishes at first order in the ratio of @xmath118 , where @xmath119 is the minor radius of the torus ( and @xmath30 corresponds the the major radius ) .
therefore , the first non - zero perturbation must occur at second or higher order .
here we have shown that for our thin tube model the first non - zero perturbation to the frequency does indeed occur at second order in the radius of curvature , as a simple dimensional analysis would indicate .
many of the observed transverse oscillations appear to be horizontal ( e.g. , * ? ? ?
* ) , and therefore in the binormal direction . unfortunately , our assumption of an isolated magnetic loop is particularly suspect for such oscillations .
most of what we call coronal loops are probably just a bundle of field lines that are part of a much larger magnetic structure like a magnetic arcade .
we identify this particular bundle as a
coronal loop " solely because it is selectively illuminated by a local heating process .
if the loop is part of a magnetic arcade , oscillations in the direction of the principle normal ( sometimes called vertical " oscillations ) may be unaffected by the larger arcade structure , because the undulations do not impinge on neighboring arcade field lines .
however , binormal ( horizontal ) oscillations must inherently impact nearby field lines and the anisotropic nature of the arcade structure must modify wave propagation . in particular , waves may no longer be confined to the loop .
spanwise propagation along the arcade is possible , and since the other loops in the arcade may not be illuminated , this propagation may go undetected .
such waves may still be trapped , but the problem is inherently two - dimensional and our one - dimensional analysis incorrect . despite these difficulties , there are distinct differences between the two polarizations of waves that may be observable .
unfortunately , the eigenfunctions are of little direct use because the two polarizations have identical eigenfunctions ( to second order in the square of the magnetic bond number ) .
therefore , observations of loop motion must be able to directly account for projection and occultation effects if they are to determine polarization .
the frequencies of oscillation are not degenerate , however , and offer promise to winnow between polarization modes .
many observational and theoretical studies have examined the diagnostic value of the ratio of eigenfrequencies ( for a review see * ? ? ?
* ) , in particular the ratio of the fundamental and first overtone . for a straight magnetic fibril in the absence of gravity and stratification ,
the ratio of the first overtone to the fundamental @xmath120 is 2 .
observational differences from 2 have been interpreted as evidence for wave speed stratification ( e.g. , * ? ? ?
* ) , and potential wave speed models have been rejected based on whether the model generates a ratio tha is less than or greater than 2 .
as we shall soon see , curvature and buoyancy forces can also cause the ratio to shift from the canonical value of 2 , without any spatial variation in the wave speed . for our model ,
the frequency ratio for any given pair of neighboring modes is given by @xmath121 \ ; , \ ] ] where the upper sign ( + ) corresponds to oscillations in the normal direction and the lower sign ( @xmath122 ) to those in the binormal direction .
for the ratio of the first overtone to the fundamental , this becomes @xmath123 since the effects of the variation in the loop length are dispersionless , they have dropped out in these ratio expressions , leaving just the effects of buoyancy and curvature .
direct buoyant augmentation of the restoring force causes the ratio to decrease , while curvature of the loop causes the ratio to increase .
since the normal polarization senses the effects of loop curvature , whereas the binormal polarization does not , the ratio of the first overtone to the fundamental is greater than 2 for normal oscillations and less than 2 for binormal displacements .
the diagnostic value of the frequency ratio is clearly predicated on the knowledge of which polarization one is observing .
while we have assumed that the coronal loop is embedded in a field - free atmosphere , we can speculate what the effects of an external magnetic field might be .
let s first consider the equilibrium .
the inclusion of an external magnetic field only provides minor modifications to the steady state form of the equation of motion ( [ eqn : spruit ] ) .
the tangential component , equation ( [ eqn : equil_s ] ) , remains unchanged as it depends only on internal fluid properties .
the same can be argued for the magnetic tension as long as the thin - tube approximation holds .
the buoyancy force is modified , however , because the density difference between the internal and external fluid is a function of the external magnetic field through equilibration of the total pressure across the tube . however ,
if the external field is force - free , the density scale height of the external fluid is unaffected and density changes only manifest as a different value of the magnetic bond number @xmath0see equation ( [ eqn : epsilon ] ) .
therefore , the equation that describes the equilibrium shape of the loop , equation ( [ eqn : eqode ] ) , remains valid in the presence of external field , and of course the solutions of that equation also remain valid . the existence of an external magnetic field is not as benign for the wave equations satisfied by the kink waves .
the presence of the surrounding magnetic field can be felt by the waves through backreaction forces caused by the tube s motion .
the well - known effect of enhanced inertia , which accounts for the fact that a moving tube must move external fluid out of its way , has already been included in our equation of motion ( [ eqn : spruit ] ) .
however , when the external fluid is magnetized , there is an additional effect because bending of the tube requires bending of the external field as well .
therefore , the external field @xmath124 provides additional stiffness that increases the kink wave speed ,
@xmath125 this well - known expression for the kink wave speed ( e.g. , * ? ? ?
* ) displays both backreaction forces explicitly . in the denominator
the sum of the internal and external densities is the effect of the enhanced inertia , while in the numerator the appearance of @xmath126 is the result of the increases stiffness .
these forces are derived and identified explicitly in appendix [ app : backreaction ] for a thin tube in the absence of gravity .
when including the effects of an external magnetic field , one is tempted to simply replace the kink speed @xmath36 appearing in the wave equations ( [ eqn : xieqn_raw ] ) and ( [ eqn : zetaeqn_raw ] ) with the proper wave speed @xmath127 , @xmath128 from the motivating calculation in appendix [ app : backreaction ] it is evident that such a substitution is justified in the absence of curvature of the loop .
when curvature is present , the validity of the substitution seems physically appropriate , but remains mathematically unproven .
while these two wave equations possess the same form as those solved previously , equations ( [ eqn : xieqn_raw ] ) and ( [ eqn : zetaeqn_raw ] ) , there is one subtle difference .
the ratio appearing in front of the first derivative in each equation is no longer related to the parameter @xmath0 in the same fashion .
previously , the denominator depended on @xmath36 instead of @xmath127 .
since , the magnetic tension in the equilibrium still depends on @xmath36 ( and not @xmath127 ) , after inserting @xmath88 appropriate for the equilibirum , we find @xmath129 with @xmath130 being a factor that depends on the plasma parameter within the tube , @xmath71 , and outside the tube , @xmath131 , @xmath132 since the corona is magnetically dominated ( i.e. , @xmath133 and @xmath134 ) , @xmath130 is generally quite close to a value of one - half .
the form of our previous wave equations can be recovered only for @xmath135 . while this seems like a minor difference , the change of variable that we performed to remove the first derivatives in equations ( [ eqn : xieqn ] ) and ( [ eqn : zetaeqn ] ) is predicated on @xmath135 .
so , while the basic form of the wave equations remains unchanged when we include an external magnetic field , the detailed perturbation analysis performed in
[ sec : kinkwaves ] is no longer valid .
it would be relatively trivial to repeat the perturbation analysis for this new case , but we choose to leave such an effort to a future study .
we have developed a model of a coronal loop that self - consistently includes the effects of gravity and curvature .
this has been accomplished for a loop that is treated as an isolated magnetic fibril embedded in an isothermal field - free corona .
for such a model , the shape adopted by the loop when in a state of static equilibrium is completely determined by a balance between the forces of buoyancy and magnetic tension .
the height of the loop depends on the magnetic bond number @xmath0 which is usually small for coronal loops . for these small values ,
the loops are low - lying with weak curvature . in the other extreme ,
no stable equilibrium exists for @xmath136 .
this stability criterion is actually a restriction on the footpoint separation compared to the coronal density scale height .
loops with a footpoint separation greater than @xmath137 are unstable . since the static loop is confined within a vertical plane , and only possesses curvature within this plane , the two polarizations of transverse oscillations ( those in the direction of the principle normal and those in the direction of the binormal ) decouple .
the wave equations that describe these oscillations both have restoring forces from buoyancy and from the magnetic tension induced by undulations . the equation that describes the normal polarization also possesses a tension term that arises from the curvature of the loop itself .
the effects of the restoring force of buoyancy and of background curvature both enter the eigenfrequencies at second order in the magnetic bond number .
buoyancy increases the frequency and curvature reduces it .
the ratio of the frequencies of two eigenmodes of neighboring order @xmath138 differ from the canonical value of @xmath139 that would be achieved for a straight loop in the absence of gravity with constant kink speed .
oscillations confined to the plane of the loop , or those polarized in the normal direction , have a frequency ratio that exceeds this canonical value , @xmath140 , whereas horizontal oscillations , or the binormal polarization , have a ratio that is less than the canonical value , @xmath141 .
these conditions on the frequency ratios are achieved for a loop with a constant kink wave speed , independent of the pressure scale height .
this work is supported by nasa , msrc ( university of sheffield ) and stfc ( uk ) .
bwh acknowledges nasa grants nnx08aj08 g , nnx08aq28 g , and nnx09ab04 g .
in order to see how the backreaction forces might appear for a magnetic fibril moving through a _ magnetized _ external fluid , we perform a simple wave emission calculation for a field configuration amenable to analytical treatment and direct identification of the relevant forces .
following @xcite , consider a magnetized cylinder embedded in a magnetized atmosphere in the absence of gravity . the magnetic field within the cylinder is constant and points in the @xmath17-direction , @xmath142 .
the magnetic field in the region exterior to the cylinder is also constant and points in the same direction @xmath143 .
while the two fields are aligned , they may potentially differ in strength .
the mass densities inside and outside the cylinder are @xmath11 and @xmath12 , respectively .
magnetosonic waves outside the cylinder have a simple analytic form when expressed in polar coordinates with the axis of the coordinate system coaligned with the axis of the cylinder .
a general form for the radial fluid displacement @xmath144 and the total pressure perturbation @xmath145 are as follows @xcite : in these equations , @xmath148 , @xmath149 and @xmath150 correspond to the external values of the sound , alfvn , and slow wave speeds , respectively .
the axial and radial wavenumbers are denoted @xmath151 and @xmath152 , and the temporal frequency is @xmath89 .
the function @xmath153 is a @xmath154th - order bessel function or hankel function .
primes denote differentiation with respect to the argument of the bessel function
. the radial displacement of the external fluid must match the normal motion of the tube s outer surface .
therefore , the wavefield in the external medium must have the same @xmath160 azimuthal dependence as appears in equation ( [ eqn : xi_r ] ) and will be composed of only the dipole components ( @xmath161 ) .
further if @xmath162 , causality requires that the wavefield only possesses outward propagating waves , leading to the following form : with the function @xmath164 corresponding to an outward propagating hankel function of the first kind , @xmath165 , and the wave amplitude @xmath166 fixed by the displacement matching condition at the tube s surface , the backreaction forces acting on the tube due to the tube s motion through the external fluid can be derived by integrating the forces acting on the tube over the tube s cross - section . in the absence of gravity
the only forces acting on the tube are magnetic tension and total pressure .
if the tube is thin , we may ignore internal variations of the magnetic field strength @xmath13 and the cross - sectional mean of the tension force acting on the tube is given by the cross - sectional mean of the total pressure force can be expressed as the integral of the total pressure perturbation around the circumference of the tube by using stokes theorem to convert the area integral into a contour integral , the integral appearing in this equation is trivial because the total pressure field on the surface of the tube only has dipole dependence .
using equations ( [ eqn : pi_e ] ) and ( [ eqn : acoef ] ) this reduces to we can immediately see that there are two backreaction forces .
both arise from the total pressure in the external fluid .
the term proportional to @xmath175 is the familiar effect of enhanced inertia and arises from the @xmath176 factor appearing in equation ( [ eqn : pi_e ] ) . in the absence of external field , this would be the only term .
however , when external field is present , the total pressure in the external fluid is reduced by the fractional amount @xmath177 compared to pressure in the unmagnetized fluid .
the term proportional to the square of the alfvn speed leads to an additional backreaction term @xmath178 that represents the force needed to bend the external field lines when the tube itself bends .
the external field provides additional stiffness that increases the wave speed . | loops of magnetic field in the corona are observed to oscillate and these oscillations have been posited to be the superposition of resonant kink waves , trapped between the loop s footpoints in the photosphere . to date , most analyses of these oscillations have concentrated on calculating the frequency shifts that result from spatial variation in the kink wave speed ( produced primarily by stratification in the density and magnetic field strength ) .
further , most have ignored gravity and treated the loop as a straight tube . here
we ignore spatial variation in the wave speed , but self - consistently include the effects of gravity and loop curvature in both the equilibrium loop model and in the wave equation that governs the propagation of the kink waves that live upon the loop .
we model a coronal loop as an isolated , thin , magnetic fibril that is anchored at two points in the photosphere .
the equilibrium shape of the loop is determined by a balance between magnetic buoyancy and magnetic tension , which is characterized by a magnetic bond number @xmath0 that is typically small @xmath1 .
this balance produces a loop that has a variable radius of curvature , with the legs being relatively straight and the apex of the loop the most curved .
further , a loop becomes unstable to buoyant rise if the footpoint separation becomes larger than @xmath2 times the coronal pressure scale height .
the resonant kink waves of such a loop come in two polarizations that are decoupled from each other : waves with motion completely within the plane of the loop ( normal oscillations or
vertical modes " ) and waves with motions that are completely horizontal , perpendicular to the plane of the loop ( binormal oscillations or horizonal modes " ) .
we solve for the eigensolutions of both polarizations using perturbation theory for small magnetic bond number . for modes of the same order , normal oscillations have smaller eigenfrequencies than binormal oscillations .
the additional forces of buoyancy and magnetic tension from the curvature of the loop increase and decrease the mode frequencies , respectively .
the ratio of the frequencies of the first overtone to the fundamental mode a common diagnostic used to assess the stratification of the wave speed along the loop is modified by the inclusion of buoyancy and curvature .
we find that the normal polarization possesses a frequency ratio that exceeds the canonical value of 2 , whereas the binormal polarization has a ratio less than 2 . |
in this paper , we consider linear evolution equations of the form @xmath5 where the ( possibly unbounded ) operators @xmath6 , @xmath7 and @xmath8 generate @xmath9 semi - groups for positive @xmath10 over a finite or infinite banach space @xmath11
. equations of this form are encountered in the context of _ parabolic _ partial differential equations , a prominent example being the inhomogeneous _ heat equation _
@xmath12 where @xmath13 , @xmath14 or @xmath15 and @xmath16 denotes the laplacian in @xmath17 . a method of choice for solving numerically ( [ eq : para ] )
consists in advancing the solution alternatively along the exact ( or numerical ) solutions of the two problems @xmath18 upon using an appropriate sequence of steps , high - order approximations can be obtained -for instance with exact flows- as @xmath19 the simplest example within this class is the _ lie - trotter splitting _ @xmath20 which is a first order approximation to the solution of ( [ eq : para ] ) , while the _ symmetrized _ version @xmath21 is referred to as _
strang splitting _ and is an approximation of order @xmath22 .
the application of splitting methods to evolutionary partial differential equations of parabolic or mixed hyperbolic - parabolic type constitutes a very active field of research . for this class of problems it makes sense to split the spatial differential operator , each part corresponding to a different physical contribution ( e.g. , reaction and diffusion ) .
although the formal analysis of splitting methods in this setting can be carried out by power series expansions ( as in the case of ordinary differential equations ) , several fundamental difficulties arise , however , when establishing convergence and rigorous error bounds for unbounded operators @xcite .
partial results exist for hyperbolic problems @xcite , parabolic problems @xcite and for the schrdinger equation @xcite , even for high order splitting methods @xcite . in @xcite
, it has been established that , under the two conditions stated below , a splitting method of the form ( [ eq : splittingmethod ] ) is of order @xmath23 for problem ( [ eq : para ] ) if and only if it is of order @xmath23 for _ ordinary differential equations _ in finite dimension . in other words ,
if and only if the difference @xmath24 admits a formal expansion of the form @xmath25 the two referred conditions write ( see @xcite for a complete exposition ) : 1 .
_ semi - group property _ : @xmath6 , @xmath7 and @xmath8 generate @xmath9 semi - groups on @xmath11 and , for all positive @xmath10 , @xmath26 for some positive constants @xmath27 , @xmath28 and @xmath29 .
2 . _ smoothness property _ : for any pair of multi - indices @xmath30 and @xmath31 with @xmath32 , and for all @xmath33 $ ] , @xmath34 for a positive constant @xmath35 . however , designing high - order splitting methods for ( [ eq : para ] ) is not as straightforward as it might seem at first glance . as a matter of fact ,
the operators @xmath6 and @xmath7 are only assumed to generate @xmath9 semi - groups ( and not groups ) .
this means in particular that the flows @xmath36 and/or @xmath37 may not be defined for negative times ( this is indeed the case , for instance , for the laplacian operator ) and this prevents the use of methods which embed negative coefficients .
given that splitting methods with real coefficients must have some of their coefficients @xmath38 and @xmath39 negative to achieve order @xmath0 or more , this seems to indicate , as it has been believed for a long time within the numerical analysis community , that it is only possible to apply exponential splitting methods of at most order @xmath40 . in order to circumvent this order - barrier ,
the papers @xcite and @xcite simultaneously introduced _ complex - valued _ coefficients with positive real parts .
it can indeed be checked in many situations that the propagators @xmath41 and @xmath42 are still well - defined in a reasonable distribution sense for @xmath43 , provided that @xmath44 . using this extension from the real line to the complex plane , the authors of @xcite and @xcite built up methods of orders @xmath0 to @xmath1 by considering a technique known as _ triple - jump composition _ and made popular by a series of authors : creutz & gocksch @xcite , forest @xcite , suzuki @xcite and yoshida @xcite .
+ in this work , we continue the search for new methods with complex coefficients with positive real parts .
eventually , our objective is to show that , compared to the methods built in @xcite and @xcite by applying the triple - jump ( or quadruple - jump ) procedure , it is possible to construct more efficient methods and also of higher order by solving directly the polynomial equations arising from the order conditions . in particular , we construct methods of order @xmath2 and @xmath3 with minimal local error constants among the methods with minimal number of stages .
we also construct a method of order @xmath4 , obtained as a composition based on an appropriate 8th order splitting method .
here we are particularly interested in obtaining new splitting methods for reaction - diffusion equations .
these constitute mathematical models that describe how the population of one or several species distributed in space evolves under the action of two concurrent phenomena : _ reaction _ between species in which predators eat preys and _ diffusion _ , which makes the species to spread out in space . from a mathematical point of view , they belong to the class of semi - linear parabolic partial differential equations and
can be represented in the general form @xmath45 where each component of the vector @xmath46 represents the population of one species , @xmath47 is the real diagonal matrix of diffusion coefficients and @xmath48 accounts for all local interactions between species .
yields fisher s equation and is used to describe the spreading of biological populations ; the choice @xmath49 describes rayleigh benard convection ; the choice @xmath50 with @xmath51 arises in combustion theory and is referred to as zeldovich equation . ] strictly speaking , the theoretical framework introduced in @xcite does not cover this situation if @xmath48 is nonlinear , so that ( apart from section [ sect : nt ] , where we successfully integrate numerically an example with nonlinear @xmath48 ) we will think of @xmath48 as being linear .
the important feature of @xmath52 here is that it has a * real * spectrum : hence , any splitting method involving complex steps with positive real part is suitable for that class of problems . in principle
, one could even consider splitting methods with @xmath38 s having positive real part and unconstrained complex @xmath39 s .
it may be worth mentioning that the size of the arguments of the @xmath38 coefficients of the splitting method is a critical factor when the diffusion operator involves a complex number , for instance , an equation of the form @xmath53 where @xmath54 is a complex number with @xmath55 .
the choice @xmath56 is known as the cubic ginzburg
landau equation @xcite .
in this situation , the values of the @xmath57 determine whether the splitting method makes sense for this specific value of @xmath54 .
if for all @xmath58 , @xmath59 $ ] then the method is well defined . in order for the method to be applicable to such class of equations , it would make sense trying to minimize the value of @xmath60 . in this work , however , we focus on the case where the operator @xmath6 has a real spectrum , and thus we will only require that @xmath61 ( i.e. , @xmath62 ) while minimizing the local error coefficients . in this sense , we have observed that the coefficients of accurate splitting methods with @xmath62 tend to have also @xmath39 coefficients with positive real parts . the plan of the paper is the following . in section [
sect : aob ] , we shall prove that if an @xmath63-jump construction is carried out from a basic symmetric second - order method , it is bounded to order @xmath1 and no more and we will further justify why solving directly the system of order conditions leads to more efficient methods . in section [
sect : cmw ] , we solve the corresponding order conditions of methods based on compositions of second order integrators and construct several splitting methods whose coefficients have positive real part . in particular , in subsection
[ ssect : comp ] we present splitting methods of orders @xmath2 and @xmath3 , obtained as a composition scheme with strang splitting as basic integrator . in addition , with the aim of showing that 14 is not an order barrier in general , we have built explicitly a method of order @xmath4 as a composition of methods of order 8 , which in turn is obtained by composing the second order strang splitting . in section [
sect : nt ] we describe the implementation of the various methods obtained in this paper and show their efficiency as compared to already available methods on two test problems .
finally , section [ sec.6 ] contains some discussion and concluding remarks .
a simple and very fruitful technique to build high - order methods is to consider compositions of low - order ones with fractional time steps . in this way , numerical integrators of arbitrarily
high order can be obtained . for splitting methods aimed to integrate problems of the form ( [ eq : para ] ) , it is necessary , however , that the coefficients have positive real part . the procedure has been carried out in @xcite , where composition methods up to order 14 have been constructed .
we shall prove here that 14 constitutes indeed an order barrier for this kind of approach .
in other words , the composition technique used in @xcite does not allow for the construction of methods having all their coefficients in @xmath64 with orders strictly greater than @xmath1 . with this goal in mind we consider the following two families of methods : + given a method of order @xmath23 , @xmath65}(h ) = { \ensuremath{\mathrm{e}}}^{h(a+b)}+\mathcal{o}(h^{p+1})$ ] , a sequence of methods of orders @xmath66 can be defined recursively by the compositions @xmath67}(h)=\prod_{i=1}^{m_q }
\phi^{[p+q-1]}(\alpha_{q , i } h ) : = \phi^{[p+q-1]}(\alpha_{q,1 } h ) \ , \cdots \
\phi^{[p+q-1]}(\alpha_{q , m_{q } } h),\ ] ] where for all @xmath68 , @xmath69 ( hereafter , we will interpret the product symbol from left to right ) .
notice that if @xmath70 is even , the second condition has only complex solutions .
+ given a * symmetric * method of order @xmath71,@xmath72}(h)$ ] , a sequence of methods of orders @xmath73 , @xmath74 can be defined recursively by the * symmetric * compositions @xmath75}(h)=\prod_{i=1}^{m_q } \tilde{\phi}^{[2(p+q)-2]}(\alpha_{q , i } h)\ ] ] where @xmath76 , and for all @xmath68 , @xmath77 methods of this class with real coefficients have been constructed by creutz & gocksch @xcite , suzuki @xcite and yoshida @xcite .
however , the second condition clearly indicates that at least one of the coefficients must be negative .
in contrast , there exist many complex solutions with coefficients in @xmath78 . generally speaking , starting from @xmath79}(h)$ ] ,
the @xmath80-th member of family i is of the form @xmath81}(h)=
\prod_{i_p=1}^{m_p } \left ( \prod_{i_{p-1}=1}^{m_{p-1 } } \left ( \ldots \left ( \prod_{i_{1}=1}^{m_1 } \phi^{[1]}(\alpha_{p , i_{p}}\alpha_{p-1,i_{p-1 } } \cdots\alpha_{1,i_{1}}h ) \right ) \ldots \right ) \right)\ ] ] and has coefficients @xmath82 a similar expression holds , of course , for methods of family ii , starting from @xmath83}(h)$ ] .
symmetric compositions for the cases @xmath84 and @xmath85 , correspond to the triple and quadruple jump techniques , respectively . [ lemcons ]
let @xmath86 be a consistent method ( i.e. , a method of order @xmath87 ) and assume that the method @xmath88 is also consistent ( i.e. , @xmath89 ) .
if there exists @xmath90 , @xmath91 , such that @xmath92 , then any consistent method of the form @xmath93 has at least one coefficient @xmath94 , @xmath95 , such that @xmath96 . by consistency one has @xmath97 , so that @xmath98 this implies the statement .
[ lemsect ] for @xmath99 and @xmath100 , consider @xmath101 such that @xmath102 , @xmath103 for @xmath104 .
then we have @xmath105 all the @xmath106 s belong to the sector @xmath107 with ( -1,-3)(5,4 ) ( 4;50)(0,0)(4;10 ) ( 0,0)(3.5,-2 . ) ( 2.9,1.9)@xmath108 ( 3.7,0.4)@xmath109 ( 3.1,-0.6)@xmath108 ( -1,0)(4,0 ) ( 0,-3)(0,3 ) ( -0.1,3.3)@xmath110 ( 4.3,0.2)@xmath111 ( 2.1,0.6)@xmath112 ( 1.2,1.1)@xmath113 ( 1.1,-0.2)@xmath114 ( 2.0,-0.7)@xmath115 ( 2.,0.9)@xmath116 ( 1.1,1.45)@xmath116 ( 1.5,-0.2)@xmath116 ( 1.55,-0.8)@xmath116 ( -2,-3)(5,4 ) ( 0,0)(-0.69,3.94 ) ( 0,0)(1.99,-3.46 ) ( 0,0)(3.75,1.35 ) ( 2.9,2.2)@xmath117 ( 3.7,0.4)@xmath109 ( 3.1,-0.9)@xmath117 ( -1,0)(4,0 ) ( 0,-3)(0,3.5 ) ( -0.1,3.8)@xmath110 ( 4.3,0.2)@xmath111 ( 3.6,3.52)@xmath118 ( -1,3.25)@xmath119 ( 1.8,-0.6)@xmath120 ( 2.,-2.65)@xmath121 ( 3.19,3.6)@xmath116 ( -.56,3.25)@xmath116 ( 2.21,-0.6)@xmath116 ( 1.6,-2.65)@xmath116 @xmath122 where @xmath123 is the principal value of the argument ( see the left picture of figure [ fig : envsect ] ) . now , assume that @xmath124 . then , all the @xmath125 s belong to @xmath126 , which , since @xmath127 , is a convex set .
this implies that @xmath128 also belongs to @xmath126 ( see the right picture of figure [ fig : envsect ] ) .
the inequality @xmath127 being strict and the @xmath106 s non - zero , we have furthermore @xmath129 , which contradicts the assumption .
the result follows .
[ theorem2.1a ] ( i ) starting from a first - order method @xmath79}(h)$ ] , all methods @xmath130}(h)$ ] of order @xmath131 , @xmath132 from family i have at least one coefficient with negative real part . (
ii ) starting from a second - order symmetric method @xmath83}(h)$ ] , all methods @xmath133}(h)$ ] of order @xmath134 , @xmath135 from family ii have at least one coefficient with negative real part .
we prove at once the two assertions .
we first notice that , according to lemma [ lemcons ] , if method @xmath65}(h)$ ] of family i ( respectively , method @xmath72}(h)$ ] of family ii ) , has a coefficient with negative real part , then all subsequent methods @xmath136}(h)$ ] , @xmath68 , of family i ( respectively , methods @xmath137}(h)$ ] of family ii ) , also have a coefficient with negative real part .
hence , we assume that all methods @xmath138}(h)$ ] , @xmath139 from family i ( respectively , all methods @xmath140}(h)$ ] of family ii ) , have all their coefficients in @xmath141 .
using lemma [ lemsect ] we have @xmath142 ( respectively @xmath143 so that @xmath144 ( respectively @xmath145 now , since @xmath146 , @xmath147 in the first case and thus the first statement follows . for family ii , since @xmath148 , then @xmath149 , thus leading to the second statement .
no method of family i with coefficients in @xmath141 can have an order strictly greater than @xmath0 .
such methods of order @xmath0 have been constructed in @xcite .
similarly , no method of family ii with coefficients in @xmath141 can have an order strictly greater than @xmath1 .
such methods with orders up to @xmath1 have been constructed in @xcite .
for instance , let us consider the quadruple jump composition @xmath150}(h)= \tilde{\phi}^{[2p]}(\alpha_{p,1}h ) \ , \tilde{\phi}^{[2p]}(\alpha_{p,2}h ) \ , \tilde{\phi}^{[2p]}(\alpha_{p,2}h ) \
, \tilde{\phi}^{[2p]}(\alpha_{p,1}h),\ ] ] where @xmath151 and @xmath152 .
this system has @xmath23 solutions ( and their complex conjugate ) .
the solution with minimal argument is , as noticed in @xcite , @xmath153 ( and its complex conjugate ) .
it is straightforward to verify that @xmath154 and @xmath155 , so that @xmath156 comparing with the proof of theorem [ theorem2.1a ] , we observe that the bounds obtained there are sharp since a family of methods do exist satisfying the equality .
it _ is _ however possible to construct a composition method with all coefficients having positive real part of order strictly greater than @xmath1 directly from a symmetric second order method .
for example , in subsection [ ssect : comp ] we present a new method of sixteenth - order built as @xmath157}(h)= \prod_{i=1}^{21 } \phi^{[8]}(\alpha_{i}h ) , \qquad \mbox{with } \qquad \phi^{[8]}(h)=\prod_{j=1}^{15 } \phi^{[2]}(\beta_{j}h)\ ] ] and the coefficients satisfying @xmath158 for all @xmath159 , @xmath160 , with @xmath161 . here ,
@xmath162}(h)$ ] is a symmetric composition of symmetric second order methods , but it is not a composition of methods of order 4 or 6 , and similarly for @xmath163}(h)$ ] , which is not a composition of methods of orders @xmath164 , @xmath165 or @xmath1 .
[ theorem2.4 ] splitting methods of the class ( [ eq : splittingmethod ] ) with @xmath62 exist at least up to order 44 . in @xcite ,
a fourth - order method was obtained with @xmath166 . in a similar way
, we have also built a sixth - order scheme with @xmath166 ( whose coefficients can be found at ` http://www.gicas.uji.es/research/splitting-complex.html ` ) . using this as the basic method in the quadruple jump ( [ eq : qjump ] ) with coefficients chosen with the minimal argument and , since @xmath167 we conclude that all methods obtained up to order 44 will satisfy that @xmath62 .
the question of the existence of splitting schemes at any order with @xmath62 remains still open .
we have seen that the composition technique to construct high order methods inevitably leads to an order barrier . in addition , the resulting methods require a large number of evaluations ( i.e. @xmath168 or @xmath169 evaluations to get order @xmath170 using the triple or quadruple jump , respectively ) and usually have large truncation errors . in this section
we show that , as with real coefficients , it is indeed possible to build very efficient high order splitting methods whose coefficients have positive real part by solving directly the order conditions necessary to achieve a given order @xmath23 . these are , roughly speaking , large systems of polynomial equations in the coefficients @xmath38 , @xmath39 of the method ( [ eq : splittingmethod ] ) , arising by requiring that the formal expansion of the method satisfies ( [ eq : formalle ] ) for arbitrary non - commuting operators @xmath6 and @xmath7 .
different ( but equivalent ) formulations of such order conditions exist in the literature @xcite . among them ,
the one using lyndon multi - indices is particularly appealing .
it was first introduced in @xcite ( see also @xcite ) and can be considered as a variant of the classical treatment obtained in @xcite .
this analysis shows that the number of order conditions for general splitting methods of the form ( [ eq : splittingmethod ] ) grows very rapidly with the order @xmath23 , even when one considers only symmetric methods .
for instance , there are @xmath171 independent 8th - order conditions and @xmath172 10th - order conditions for a consistent symmetric splitting method .
it makes sense , then , to examine alternatives to achieve order higher than six .
this can be accomplished by taking compositions of a basic symmetric method of even order . in particular ,
if we consider any of the two versions of strang splitting ( [ eq : strang2 ] ) as the basic method @xmath173 , then , for each @xmath174 , @xmath175 will be a new splitting method of the form ( [ eq : splittingmethod ] ) .
now the consistency condition reads @xmath176 as for the additional conditions to attain order @xmath23 , these can be obtained by generalizing the treatment done in @xcite .
splitting methods with very high order can be constructed as ( [ eq : comppsi ] ) by considering as basic method @xmath173 a symmetric method of even order @xmath177 . in that case
, it can be shown that the corresponding number of order conditions is considerably reduced with respect to ( [ eq : splittingmethod ] ) .
thus , for instance , if @xmath173 is a symmetric splitting method of order eight , a method ( [ eq : comppsi ] ) satisfying the consistency condition ( [ eq : consistencygamma ] ) and the symmetry condition @xmath178 only needs to satisfy 10 additional conditions to attain order sixteen . to construct splitting methods of a given order @xmath23 within a family of schemes
, we choose the number @xmath63 of stages in such a way that the number of unknowns equals the number of order conditions , so that one typically has a finite number of isolated ( real or complex ) solutions , each of them leading to a different splitting method . among them
, we will be interested in methods such that , either @xmath179 ( and each @xmath39 are arbitrary complex numbers ) or @xmath62 and @xmath180 .
the relevant question at this point is how to choose the ` best ' solution in the set of all solutions satisfying the required conditions .
it is generally accepted that good splitting methods must have small coefficients @xmath181 .
methods with large coefficients tend to show bad performance in general , which is particularly true when relatively long time - steps @xmath182 are used .
in addition , as mentioned in the introduction , when applying splitting methods to the class of problems considered here , the arguments of the complex coefficients @xmath181 must also be taken into account . in order to chose
the best scheme among two methods with coefficients of similar size included in sectors with similar angle , we analyze the leading term of the local error of the splitting method .
if ( [ eq : splittingmethod ] ) is of order @xmath23 , then we formally have that @xmath183 where @xmath184 is given in terms of the original coefficients @xmath38 , @xmath39 of the integrator by @xmath185 ( @xmath186 ) and each @xmath187 is a linear combination of polynomials in @xmath188 @xcite . incorporating that into the results in @xcite
, it can be shown that , if the smoothness assumption stated in the introduction is increased from @xmath131 to @xmath189 , then for sufficiently small @xmath182 , the local error is dominated by @xmath190 .
our strategy to select a suitable method among all possible choices is then the following : first choose a subset of solutions with reasonably small maximum norm of the coefficient vector @xmath191 , and then , among them , choose the one that minimizes the norm @xmath192 of the coefficients of the leading term of the local error .
this seems reasonable if one is interested in choosing a splitting method that works fine for arbitrary operators @xmath6 and @xmath7 satisfying the semi - group and smoothness conditions mentioned in the introduction .
of course , this does not guarantee that a method with a smaller value of ( [ eq : lenorm ] ) will be more precise for any @xmath6 and @xmath7 than another method with a larger value of ( [ eq : lenorm ] ) .
however , we have observed in practice when solving the order conditions of different families of splitting methods , that the solution that minimizes ( [ eq : lenorm ] ) tend to have smaller values of most ( or even all ) coefficients @xmath193 when compared to a solution having a larger norm ( [ eq : lenorm ] ) of the coefficients of the leading term of the local error .
when @xmath6 and @xmath7 are operators in a real banach space @xmath11 , then it makes sense to compute the approximations @xmath194 , as @xmath195 . in that case , the argument above holds with @xmath196 replaced by @xmath197 and the local error coefficients @xmath198 replaced by @xmath199 . in that case ,
( [ eq : lenorm ] ) should be replaced by @xmath200 as a general measure of leading term of the local error .
[ [ order-6 . ] ] order 6 .
+ + + + + + + + we first consider sixth - order symmetric splitting methods obtained as a composition ( [ eq : comppsi ] ) of the strang splitting ( [ eq : strang2 ] ) as basic method . in this case
the coefficients @xmath201 must satisfy three order conditions , in addition to the symmetry ( [ eq : gammasym ] ) and consistency requirements , to achieve order six .
we thus take @xmath202 , so that we have three equations and three unknowns .
such a system of polynomial equations has 39 solutions in the complex domain ( three real solutions among them ) , 12 of them giving a splitting method with coefficients of positive real part . according to the criteria described in subsection [ ssect : le ]
, we arrive at the scheme @xmath203 this method turns out to correspond to one of those obtained by chambers ( see table 4 in @xcite ) .
[ [ order-8 . ] ] order 8 .
+ + + + + + + + for consistent symmetric methods ( [ eq : comppsi ] ) of order eight , we have seven order conditions . by taking @xmath204 stages ,
one ends up with a system of seven polynomial equations and seven unknowns .
we have performed an extensive numerical search of solutions with small norm , finding 326 complex solutions . among them , 162 lead to splitting methods whose coefficients possess positive real part .
the best method , according to the criteria established in subsection [ ssect : le ] , is @xmath205 [ [ order-16 . ] ] order 16 .
+ + + + + + + + + motivated by the results in section [ sect : aob ] , we have also constructed a splitting method of order 16 .
our aim , rather than proposing a very efficient scheme , is to show that the barrier of order 14 existing for methods built by applying the recursive composition technique starting from order two ( family ii ) does not apply in general .
the construction procedure can be summarized as follows .
we consider a consistent symmetric composition of the form ( [ eq : comppsi ] ) , where now the basic method @xmath173 is any symmetric eighth order scheme . under such conditions
, ten order conditions must be satisfied to achieve order 16 .
we accordingly take @xmath206 , so that we have ten polynomial equations with ten unknowns .
we have performed an extensive numerical search of solutions with relatively small norm , finding @xmath207 complex solutions with positive real part . combined with the @xmath208 methods of order eight , this leads to @xmath209 different @xmath4-th order splitting methods with @xmath210 stages . among them , only @xmath211 give rise to splitting methods with coefficients of positive real part .
the coefficients of the method that we have determined as optimal can be found at ` http://www.gicas.uji.es/research/splitting-complex.html ` .
notice that , whereas for the sixth - order method ( [ p6s7 ] ) one has @xmath212 and then the coefficients @xmath201 are distributed in a narrower sector than for triple or quadruple jump methods , for the eighth - order method ( [ p8s15 ] ) one has @xmath213 this method , whereas being the most efficient , is not the appropriate one to be used as basic scheme to build higher order methods by composition .
the previous 16th - order integrator has been built starting with another 8th - order method whose coefficients are placed in a narrower sector .
for our numerical experiments , we consider two different test problems : a linear reaction - diffusion equation , and the semi - linear fisher s equation , both with periodic boundary conditions in space .
it should be mentioned here that this last case is not covered by the theoretical framework summarized in the introduction . for each case
, we detail the experimental setting and collect the results achieved by the different schemes .
our main purpose here is just to illustrate the performance of the new splitting methods to carry out the time integration as compared with those constructed by using the yoshida
suzuki triple jump composition technique for both examples .
notice that , in this sense , the particular scheme used to discretize in space is irrelevant .
for that reason , and to keep the treatment as simple as possible , we have applied a simple second - order finite difference scheme in space .
the numerical approximations @xmath214 obtained by each method @xmath196 are computed as @xmath215 . in other words
, we project on the real axis after completing each time step .
our first test - problem is the scalar equation in one - dimension @xmath216 with @xmath217 and periodic boundary conditions in the space domain @xmath218 $ ] .
we take @xmath219 , @xmath220 and discretize in space @xmath221 thus arriving at the differential equation @xmath222 where @xmath223 .
the laplacian @xmath16 has been approximated by the matrix @xmath6 of size @xmath224 given by @xmath225 and @xmath226 .
we take @xmath227 points and compare different composition methods by computing the corresponding approximate solution on the time interval @xmath218 $ ] . in particular , we consider the following schemes : * * strang * : the second - order symmetric strang splitting method ( [ eq : strang2 ] ) ; * * ( tj6 ) * : the sixth - order triple jump method ( proposition 2.1 in @xcite ) based on strang s second - order method ; * * ( tj6a ) * : the sixth - order triple jump method ( proposition 2.2 in @xcite ) based on strang s second - order method ; * * ( tj8a ) * : the eighth - order triple jump method ( proposition 2.2 in @xcite ) based on strang s second - order method ; * * ( p6s7 ) * : the sixth - order method ( [ p6s7 ] ) ; * * ( p8s15 ) * : the eighth - order method ( [ p8s15 ] ) .
we compute the error of the numerical solution at time @xmath228 ( in the @xmath22-norm ) as a function of the number of evaluations of the basic method ( the strang splitting ) and represent the outcome in figure [ fig : reactdiff ] . in the left panel
we collect the results achieved by the strang splitting and the previous sixth - order composition methods , whereas the right panel corresponds to eighth - order methods .
we have also included , for reference , the curve obtained by ( p6s7 ) .
the relative cost ( w.r.t .
strang ) of a method composed of @xmath63 steps is approximated by @xmath229 , where the factor @xmath230 stands here for an average ratio between the cost of complex arithmetic compared to real arithmetic .
a remarkable outcome of these experiment is that methods ( p6s7 ) and ( p8s15 ) outperform strang s splitting ( and actually all other methods tested here ) even for low tolerances .
scheme ( p8s15 ) , in particular , proves to be the most efficient in the whole range explored .
the gain with respect to triple jump methods is also significant and completely support the approach followed here .
+ our second test - problem is the scalar equation in one - dimension @xmath231 with periodic boundary conditions in the space domain @xmath218 $ ] .
we take , in particular , fisher s potential @xmath232 the splitting considered here corresponds to solving , on the one hand , the linear equation with the operator @xmath6 being the laplacian , and on the other hand , the nonlinear ordinary differential equation @xmath233 with initial condition @xmath234 note that it can be solved analytically as @xmath235 which is well defined for small complex time @xmath10 .
we proceed in the same way as for the previous linear case , starting with @xmath217 .
after discretization in space , we arrive at the differential equation @xmath236 where @xmath237 and @xmath238 is now defined by @xmath239 we choose @xmath227 and compute the error ( in the @xmath22-norm ) at the final time @xmath228 by applying the same composition methods as in the linear case .
the results are collected in figure [ fig : reactdiff-2 ] , where identical notation has been used .
notice that , strictly speaking , the theoretical framework upon which our strategy is based does not cover this nonlinear problem .
nevertheless , the results achieved are largely similar to the linear case .
in particular , the new 8th - order composition method is the most efficient even for moderate tolerances .
splitting methods with real coefficients for the numerical integration of differential equations of order higher than two have necessarily some negative coefficients .
this feature does not suppose any special impediment when the differential equation evolves in a group , but may be unacceptable when it is defined in a semi - group , as is the case with the evolution partial differential equations considered in this paper .
one way to get around this fundamental difficulty is to consider splitting schemes with complex coefficients having positive real part .
this has been recently proposed for diffusion equations in @xcite .
splitting and composition methods with complex coefficients have been considered in different contexts in the literature ( see @xcite and references therein ) . in @xcite ,
splitting methods up to order 14 with complex coefficients with non - negative real part have been recursively constructed either by the so - called triple - jump compositions or by the quadruple - jump compositions , starting from the symmetric second - order strang splitting . in this work
we prove that there exists indeed an order barrier of 14 for methods constructed in this way .
more generally , we show that no method of order higher than 14 with coefficients having non - negative real part can be constructed by sequential @xmath63-jump compositions starting from a symmetric method of order 2 .
we further show , by explicitly obtaining methods of order 16 ( as the composition of a basic symmetric method of order 8) , that this order barrier does not apply for general composition methods ( non - necessarily constructed by recursive applications of @xmath63-jump compositions ) with complex coefficients with non - negative real part .
in addition to this order barrier , another drawback of methods resulting from applying the @xmath63-jump composition procedure is that for high orders they require a larger number of stages ( i.e. number of compositions of the basic symmetric second order method ) than methods obtained by directly solving the order conditions with the minimal number of stages .
for instance , methods of order 6 ( respectively , 8) obtained with triple jump compositions need 9 ( resp .
27 ) compositions of the basic second order method , whereas , as we show in the present work , methods of order 6 ( resp . 8) can be constructed ( by directly solving for the required order conditions ) with 7 ( resp .
15 ) stages .
an analysis of the local error coefficients supported by numerical tests shows that the methods proposed here are more efficient than those obtained in @xcite by applying the recursive triple jump and quadruple jump constructions .
an additional requirement when choosing a given method is that the arguments of the complex coefficients of the scheme have also to be taken into account .
this constitutes a critical point for evolution equations where one of the operators ( say , @xmath6 ) has non - real eigenvalues in the right - hand side of the complex plane , as occurs , in particular , with the complex ginzburg landau equation ( [ eq : landau ] ) .
in such a case , splitting methods of the form ( [ eq : splittingmethod ] ) where one of the two sets of coefficients @xmath38 s or @xmath39 s is entirely contained in the positive real axis , whereas the other set is included in the right - hand side of the complex planes are particularly well suited .
such splitting methods can not be constructed as composition methods with the strang splitting as basic method , so that a separate study is required to get the most efficient schemes within this class .
based on the theoretical framework worked out in @xcite , the integrators proposed here can be applied to the numerical integration of linear evolution equations involving unbounded operators in an infinite dimensional space , like linear diffusion equations . as a matter of fact , although the theory developed in @xcite does not cover the generalization to semi - linear evolution equations , we have also included in our numerical tests a system of odes obtained from semi - linear evolution equations with a certain space discretization .
all the numerical tests carried out with periodic boundary conditions show a considerable improvement in efficiency of our new methods with respect to existing splitting schemes .
a remarkable feature of the new eighth - order composition method when applied to both the linear and semi - linear diffusion examples is that it is more efficient than all the other integrators of order @xmath240 in the whole range of tolerances explored when periodic boundary conditions are considered .
concerning other ( e.g. dirichlet and neumann ) boundary conditions , the experiments carried out in @xcite for linear problems with methods obtained by applying the triple and quadruple jump technique show the existence of an order reduction phenomenon .
its origin is attributed by the authors of @xcite to the fact that condition 2 in the introduction is not generally satisfied in this setting . in other words , terms of the form @xmath241 in ( [ eq : formalle ] )
are not uniformly bounded on the interval @xmath242 $ ] for some @xmath243 .
as a consequence , the classical convergence order is no longer guaranteed in that case .
this order reduction is also present , of course , when the methods introduced in this paper are applied to linear but also semi - linear problems . nevertheless ,
as has been pointed out in @xcite , splitting schemes of high order involving complex coefficients may still be advantageous in comparison with , say , strang splitting when applied to linear parabolic problems with dirichlet or neumann boundary conditions , as they often lead to smaller errors and the order reduction is somehow confined to a neighborhood of the boundary .
the situation , in our opinion , calls for a detailed study of the effect of boundary conditions other than the periodic case in the global efficiency of splitting methods with complex coefficients and the analysis of their applicability for more general parabolic problems than those considered in this paper .
the work of sb , fc and am has been partially supported by ministerio de ciencia e innovacin ( spain ) under the coordinated project mtm2010 - 18246-c03 ( co - financed by feder funds of the european union ) .
financial support from the `` accin integrada entre espaa y francia '' hf2008 - 0105 is also acknowledged .
am is additionally funded by project ehu08/43 ( universidad del pas vasco / euskal herriko unibertsitatea ) . | we are concerned with the numerical solution obtained by _ splitting methods _ of certain parabolic partial differential equations . splitting schemes of order higher than two with real coefficients necessarily involve negative coefficients .
it has been demonstrated that this second - order barrier can be overcome by using splitting methods with _ complex - valued _ coefficients ( with positive real parts ) . in this way ,
methods of orders @xmath0 to @xmath1 by using the suzuki
yoshida triple ( and quadruple ) jump composition procedure have been explicitly built .
here we reconsider this technique and show that it is inherently bounded to order @xmath1 and clearly sub - optimal with respect to error constants . as an alternative
, we solve directly the algebraic equations arising from the order conditions and construct methods of orders @xmath2 and @xmath3 that are the most accurate ones available at present time , even when low accuracies are desired .
we also show that , in the general case , 14 is not an order barrier for splitting methods with complex coefficients with positive real part by building explicitly a method of order @xmath4 as a composition of methods of order 8 .
+ * keywords * : composition methods , splitting methods , complex coefficients , parabolic evolution equations . + * msc numbers * : 65l05 , 65p10 , 37m15 0.8 mm |
monomethine cyanines have a distinguished role in the history of chemical science and industry.@xcite the description of their optical properties was an early target of quantum electronic structure models.@xcite the optical properties of monomethine dyes and push - pull polyenes can be described by two - diabatic - state models based on the concept of resonance between configurations with opposing charge and bond localization.@xcite .
the resonating diabatic states are shown for four typical symmetric monomethine dyes in [ fig : scheme ] .
a distinguishing feature of monomethine cyanines is the environmental sensitivity of their fluorescence emission yield.@xcite the dyes are practically non - fluorescent in fluid solutions , but can become highly fluorescent when bound to biomolecules or other constraining media.@xcite this underlies the use of monomethine cyanines as fluorescent turn - on labels in biological imaging . the fluorescence enhancement is interpreted as suppression of a bond - twisting process in the excited state , which would otherwise lead to dark twisted intramolecular ( tict ) states@xcite .
the tict states in monomethine cyanines are dark , so population of these states is associated with ultrafast loss of transient fluorescence .
the tict are characterized by a small adiabatic gap , and nonadiabatic decay can occur by conical intersections that occur at twisted geometries.@xcite enhancement of fluorescence due to suppression of excited - state twisting motion is the defining characteristic of a `` molecular rotor''@xcite .
this behavior is not limited to monomethine cyanines .
an example of a non - methine molecular rotor is the amyloid fibril - sensing dye thioflavin - t@xcite .
this molecule undergoes twisting in the excited state , allowing ultrafast internal conversion via twisted intramolecular charge - transfer ( tict ) states with small or vanishing electronic gap@xcite .
the tict states in thioflavin - t and auramine - o are dark , so that the twisting reaction is accompanied by a fast transient fluorescence decay and negligible steady - state quantum yield .
the fluorescence is enhanced upon binding to biomolecular environments or in viscous solution .
the nonradiative decay rates of thioflavin - t and the monomethine dye auramine - o have been extensively characterized by huppert and coworkers in a series of glass - forming alcohols , under conditions spanning 10 orders of magnitude in the viscosity@xcite .
these studies supported earlier indications@xcite that the nonradiative decay time could be estimated from the damped rotational stokes drift of a spherical rotor down a linear potential through an angle of one radian .
the reaction coordinate is taken to be rotation of a double bond to form a dark tict state .
a useful estimate of the rotational timescale ( which describes the non - radiative decay timsecale ) was found to be @xmath7 , where @xmath8 is the medium viscosity , @xmath9 is the effective volume of the rotor ( 20 @xmath10 were taken for thioflavin - t or auramine - o ) , and @xmath11 is the mean torque on the rotor over a rotation of one radian .
a key point was that the non - radiative decay timescale was found to depend linearly on the medium viscosity .
a linear viscosity dependence of the nonradiative decay rate was also observed in experiments on the triphenylmethane dye crystal violet , which decays via a similar mechanism.@xcite different possible situations for a one - dimensional molecular rotor system with dark tict states are illustrated schematically in figure [ fig : tictoutcomes ] . for twisting that is both favorable in the excited state and without barrier
, one expects ultrfast ( 1ps ) decay of fluorescence as the tict state is formed from the bright franck - condon ( fc ) state . provided depletion of tict excited state is not rate - limiting , which our results support ( see below ) , ground state recovery should track the fluorescence decay . for twisting that is favorable but activated ,
one expects slower fluorescence decay . again ,
if the depletion of the excited tict states is not rate - limiting , the ground state recovery should follow the fluorescence decay . when tict state formation is not favorable , we expect that they do not form on a relevant timescale and the molecule is fluorescent . if our assumption about fast depletion of the tict excited - state were wrong , so that populuation accumulates on the excited state , then the same pattern would appear in the ultrafast fluorescence , but we would expect the ground state recovery to lengthen independently . for molecular rotors with the dimensions characteristic of monomethine cyanines , the timescale for generation of a charge - transfer state by twisting and the timescale for solvation of the same charge - transfer state can not be reliably separated .
the nonradiative decay times for thioflavin - t and auramine - o in glass - forming solvents scale with the transverse debye dielectric relaxation time@xcite , but are 1 - 2 orders of magnitude smaller@xcite .
this is consistent with nonradiative decay times that are comparable to the debye _ longitudinal _ relaxation time .
the longitudinal relaxation time is the appropriate timescale to describe dielectric relaxation following a sudden change in the molecular charge density@xcite . for alcoholic solutions ,
the longitudinal relaxation time will be 10 times smaller than the transverse timescale@xcite , which is comparable to estimated twisting times .
although solvent - dependent effects can be observed at sub - ps timescales for auramine - o , these were assigned to a distinct dynamical origin.@xcite heisler and colleagues showed for auramine - o in nanoconfined water droplets that restriction of solvation by the confinement slowed the fluorescence upconversion decay.@xcite results reported by kondo and coworkers for auramine - o in bulk solutions indicated that the dominance of solvation dynamics vs reaction coordinate friction in determining the fluorescence upconversion decay could not be determined for several solvents excepting water , for which solvation dynamics was dominant.@xcite the _ n , n_-dimethylaniline rotors in auramine - o are identical to those of michler s hydrol blue in figure [ fig : scheme ] , and so the timescales for damped twisting are expected also to be the same .
if the timescales for solvation and for twisting are comparable , then a faithful description of the physics must treat these processes at a comparable level of approximation . a successful yet simple approach for the one - dimensional rotor _
n , n_-dimethylbenzoaminonitrile ( dmabn ) , which forms an emissive tict state on a timescale much longer than common for monomethine cyanines , has been to develop two - dimensional models coupled to a twisting coordinate and a single solvation coordinate that gauges the polarization state of the medium.@xcite the identification of monomethine cyanines as molecular rotors raises an interesting question , because the ground states of these molecules are a superposition of structures with different @xmath12-bond localization ( c.f .
figure [ fig : scheme ] ) . in a molecule where the double bond can be in multiple places , how does one choose the appropriate reaction coordinate to describe double bond twisting in the excited state ?
ab initio studies of the excited state surface in monomethine cyanines indicate that there are not one but _ two _ low - energy twisting channels on the excited - state potential energy surface@xcite .
general models of tict electronic structure suggest that as the bond twisting proceeds , the lowest two adiabatic states should become polarized and form a twisted intramolecular charge - transfer ( tict ) system.@xcite .
this picture is upheld by more detailed quantum chemical models@xcite . however , twisting one or the other bond will destabilize the diabatic states by different amounts because the @xmath12-bonds are in different places .
accordingly , the polarity of the tict adiabatic states will depend upon the bond that is twisted@xcite .
the dependence of the tict polarity on the identity of the twisted bond can also be explained using simple molecular orbital theories.@xcite the correlation between the path taken and the tict polarity suggests that a polar environment can distinguish the twisting pathways , and could influence their relative yields .
this has been suggested as the mechanism that underlies reversible photochromism and fluorescence switching in fluorescent proteins@xcite .
fluctuation in the charge - coordination environment have been predicted to influence the bifurcation and availability of twisting pathways in photoactive yellow protein chromophores@xcite .
modern theoretical photochemistry suggests the location , topography and energetics of conical intersection seams between adiabatic electronic states are important for understanding the mechanism and yield of photochemical reactions@xcite .
ab initio calculations and general models indicate that conical intersections can occur in monomethine dyes in configurations where one or more bonds is twisted , so that the @xmath12 orbitals on the fragments across the bond overlap weakly@xcite .
the non - adiabatic coupling ( to nuclear displacement ) diverges at conical intersections , so that internal conversion is expected to be dominated by configurations near these points . unlike transition states that separate different basins on a given surface , conical intersections are multidimensional seams embedded in the configuration space of the molecule@xcite .
although conical intersections can sometimes be discussed in terms of their minimum - energy configurations , there are cases where the relevant nonadiabatic dynamics is thought to occur far removed from the minimum - energy regions of the intersection seam@xcite .
since the twisted conical intersections in monomethine dyes arise from intersection of diabatic states with different charge localization , changes in the solvent configuration could very plausibly affect the positioning , topology and/or energetics of the seam .
the effect of polar solvation on the structure of conical intersection seams in a retinal protonated schiff base model has been studied in a series of papers by hynes and coworkers@xcite . in these papers , it was shown that solvent effects determined the placement and energy of conical intersection seams , as well as the coordinates that accessed them . a tendency for charge - transfer conical intersections to be stabilized by solvents excited - state potential in solvated systems has been observed in excited - state dynamics simulations@xcite . in this paper , we seek to develop a better intuition for how non - equilibrium solvation affects the excited - state twisting process in monomethine cyanines , with special attention to the issue that there are two possible twisting pathways that can be distinguished by the solvent .
one reason why it is important to consider simple models is that simple models make it easy to keep track of how the ideas that go into the model relate to the predictions that come out .
that way , a challenge to the model can be framed as a challenge to the underlying ideas , rather than to some detail of the model parameterization . in this paper , we focus on only a few ideas that are simple extensions of the old notion of cyanine electronic structure as a resonance between structures with opposing charge- and bond - order localization ( cf .
figure [ fig : scheme ] ) .
our model describes favorable excited - state twisting with no or a small barrier for typical monomethine cyanines in solution , leading to a manifold of twisted intramolecular charge - transfer ( tict ) configurations with small or no electronic gap .
the driving force for twisting in the isolated dye is small for dyes with absorption in the red , and increases with increasing absorption energy for a given @xmath12-bond stabilization energy .
the model predicts that solvation of products along one twisting pathway tends to destabilize the alternate pathway , so that the model can describe solvent - driven selection of distinguishable tict channels .
the position and energy of a conical intersection in the tict manifold is shown to change with the solvent configuration .
the fraction of tict states that can be made to coincide with the intersection seam grows with the solvent reorganization energy . for couplings corresponding to typical monomethine cyanines in polar solvents
, there is no minimum or `` basin '' that can be associated with an adiabatically excited tict system .
instead , the bottom of the excited state potential coincides with a conical intersection for monomethine cyanines in even moderately polar solvents . here
is how the paper is organized . in section 2
, we introduce the model , describing its physical content and parameterization .
section 3 discusses the potential energy surfaces ( pess ) given by the model .
we show that the pess of the model can be qualitatively categorized according to a phase diagram , and describe relevant features of the phase diagram .
section 4 discusses results in the context of experiments on the nonradiative decay of monomethines .
section 5 concludes .
the model hilbert space is two - dimensional , and spanned by diabatic basis states distinguished by charge and bridge @xmath12-bond order as in figure [ fig : scheme ] . in this basis
, the hamiltonian is written as ( [ eq : ham_mat ] ) .
@xmath13 the hamiltonian is comprised of two subcomponents , a dye hamiltonian @xmath14 , which is a function of the bridge - bond twist angles @xmath15 and @xmath16 , and a solvent interaction hamiltonian @xmath17 , which is a function of a single solvent configuration coordinate @xmath18 , which interpolates between equilibria with @xmath19 @xmath20 and @xmath21 @xmath22 .
the parameters @xmath23 , @xmath24 and @xmath25 are the electronic coupling strength , the @xmath12-bond stabilization energy , and the diabatic solvent reorganization energy .
the solvent interaction term @xmath17 includes a configuration - independent scalar solvation energy that depends on the refractive index @xmath26 of the medium and the born cavity radius @xmath27 characterizing the relevant terminal heterocycle .
this term is the same for both diabatic states in a symmetric dye and does not vary with the configuration of the dye or solvent ; it can be removed by resetting the zero of energy .
the transformation between the diabatic and adiabatic bases can be parameterised as a rotation by an angle @xmath28 @xmath29= \left [ \begin{array}{cc } \cos \phi & \sin \phi \\
-\sin \phi & \cos \phi \end{array } \right ] \left [ \begin{array}{c } |l\rangle\\ |r\rangle \end{array } \right].\ ] ] after resetting the zero of energy to eliminate the scalar solvation energy , the eigenvalues of the hamiltonian ( [ eq : ham_mat ] ) are @xmath30 where @xmath31 the diabatic energies vary with the twisting of the bridge bonds , through the angles @xmath15 and @xmath16 .
the dependence reflects distinguishable @xmath12-bond localization in the diabatic states in figure [ fig : scheme ] .
each diabatic state has @xmath12-bonding character localized on a distinct bridge bond .
the @xmath12-bond stabilization energy @xmath24 is the energy required to twist a @xmath12 -bond to @xmath32 , breaking the bond .
@xmath33 we are concerned with the interaction of twisting and solvation displacements , because the corresponding timescales in monomethine cyanines appear to be non - separable .
a literal interpretation of the valence - bond structures in figure [ fig : scheme ] implies also strong coupling to bond - stretching displacements .
such displacements are important for understanding linear and nonlinear spectra of these and similar systems@xcite . the off
diagonal coupling has one term representing charge transfer between rings via sequential transfers across both bridge bonds .
the direct transfer element between the ring is taken to be neglible , so that the lowest order couplings between the ( many - body ) diabatic states go as the product of transfer terms for each bond , and the modulation of transfer across each bond goes as the overlap between @xmath12-orbitals on the bonding fragments .
similar strategies have been applied to the nonlinear solvation of systems with activated tict formation along a single twisting coordinate coupled to a solvation mode .
@xmath34 the parameter @xmath23 determines the magnitude of the electronic coupling between diabatic states .
it is equal to the adiabatic gap at the ( planar ) ground state geometry .
the form of equations ( [ eq : ham_mat_elems ] ) and ( [ eq : ham_mat_od ] ) can be justified using valence bond theory@xcite , in so far as the single - particle transfer element goes as the overlap between atomic _
p _ orbitals .
the parameterization assumes a well - defined nodal plane associated with an @xmath35-hybridized monomethine bridge , so that the twist angles @xmath36 can be unambiguously defined .
the electronic coupling parameter @xmath23 is equal to the electronic gap when the dye is planar and the solvent is unpolarized , which is true at the ground state minimum . since monomethine dyes have strong , narrow absorbance lines this parameter can be estimated from absorbance spectroscopy as the energy corresponding to the band maximum .
alternatively , it can be estimated using quantum chemistry calculations . @xmath37
the absorbance of monomethine cyanines is usually sharp , and solvatochromism is negligble@xcite . for dyes such as in figure [ fig : scheme ] , with absorbance in the visible , an appropriate value for @xmath23 would be in the range 0.8 - 1.7 ev .
the @xmath12-bond energy @xmath24 is equal to the adiabatic gap if one bond is twisted and the other is planar ( in an unpolarized solvent ) .
@xmath38 this suggests the parameter @xmath24 is in - principle estimable from experiments , if transient fluorescence from the excited tict state could be collected in sufficiently non - polar solvents .
unfortunately , the transition dipole strength also vanishes at tict states , making spectroscopic estimation difficult or unfeasible in most cases . in such cases
, it may be easiest to resort to calculation of the gap using quantum chemistry computations at a suitable model of the twisted state geometry .
general considerations suggest that the parameter @xmath24 should be in the range 1.0 - 2.0 ev .
this estimate comes from the identification of @xmath24 with the energy associated with twisting a @xmath12-bond in a diabatic state with clearly defined bond alternation . to the extent that the energy of a localized double bond is transferrable
, we expect that @xmath24 should be at most comparable to the ground - state twisting energy for a molecule with definite single - double bond alternation in the adiabatic ground state .
the reason for the `` at most '' in the previous sentence is that the initial state for bond - twisting in a molecule with definite ground state bond alternation is near the minimum for the relevant diabatic state , this will not be the case in a symmetric monomethine where the ground state geometry is ambiguous .
a reasonable reference system of similar size to the molecules in figure [ fig : scheme ] might be an asymmetrically substituted stilbene .
the singlet ground state barriers for stilbene and 4-styrylpyridinium have been reported as 1.78 ev and 1.69 ev , respectively@xcite .
another route to a similar conclusion would be to note that the @xmath12-bond stabilization energy in hckel theory is just the resonance integral @xmath23 , which is usually found to be in the range 1 - 2ev@xcite . in any case
, we are led to the conclusion that an appropriate value for the barrier height will be 1 ev . for some systems , calculations of the diabatic barrier height are available , as calculations of the adiabatic gap at a suitable model tict geometry . for near - resonant models of the gfp chromophore , which are not strictly symmetric but electronically almost so , this yields a barrier height of 0.9 - 1.0ev@xcite . for the case of michler s hydrol blue
, we have an unpublished value of the twisted gap calculated using multi - state and multi - reference perturbation theory that is 1.0 ev . for a slightly truncated model of the monomethine cyanine nk88 ( cf .
figure [ fig : scheme ] ) , santoro and coworkers have calculated twisted gaps in the range 1.3 - 1.5 ev using different electronic structure approaches@xcite . these values for the diabatic barrier height @xmath23 are consistent with the arguments above . the interaction with the solvation field
is described using a model of charge transfer between conducting spheres in a linear dielectric , with cavity radii and intercavity distance chosen to reflect the geometry of a typical monomethine with modest - sized rings .
this model is well - established in the electron transfer literature @xcite .
it has also been successfully applied to model the anomalous solvatochromism of charged push - pull polyenes in solution.@xcite the diabatic solvent reorganization energy @xmath25 is calculated as the reorganization energy of charge transfer between the spheres . charged push - pull polyenes are chemically similar to monomethine dyes ( cf .
figure [ fig : scheme ] ) and are , essentially , strongly asymmetric polymethine cyanines .
the configuration of the solvent is parameterised by a single solvent coordinate @xmath18 .
those configurations which are in equilibrium with the diabatic states @xmath39 and @xmath40 ( c.f . figure [ fig : scheme ] ) we denote by @xmath41 and @xmath42 respectively .
the free energies of the diabatic states are shown in figure [ fig : nitzan ] .
the solvation interaction @xmath43 is diagonal in the diabatic basis .
the relevant matrix elements are @xmath44 where @xmath25 is the dielectric reorganization energy and @xmath45 is the born solvation energy of a charged conducting sphere of radius @xmath27 embedded in a linear dielectric ( in cgs units ) @xmath46 for a symmetric monomethine , the cavity radii will be the same , and the solvation energy terms contribute only to the trace of the hamiltonian .
they can be removed by resetting the zero of energy to @xmath47 .
the physical interpretation of the solvent reorganisation energy @xmath25 is the energy released by the non
adiabatic relaxation of the system to the minimum energy configuration of @xmath40 at @xmath41 following excitation from @xmath39 at @xmath42 , the expression for this is@xcite @xmath48 where @xmath27 is the cavity radius , @xmath49 is the distance between cavity centers , and @xmath50 and @xmath51 are the static and high - frequency dielectric constants of the medium .
we have assumed the cavity radii are the same and that the transferred charge is equivalent to the fundamental charge @xmath52 . for symmetric monomethine dyes such as shown in figure [ fig : scheme ]
, we expect @xmath53 and cavity radii @xmath54 .
table [ tab : solv_reorg_en ] shows the reorganisation energies at these values of @xmath27 and @xmath49 .
values of @xmath55 are representative of smaller endgroups , such as in dyes used as models of barrierless reaction kinetics@xcite . [ cols="<,<,<,^,^,^,^,^,^",options="header " , ] solvent reorganisation energies for a conducting - sphere model of monomethines with reasonably small rings ( like figure [ fig : scheme ] ) , as outlined in table [ tab : solv_reorg_en ] , are in excess of 1ev , close to or larger than typical values of @xmath24 , according to the last section .
huppert and coworkers have used an approximate volume of 20@xmath10 to calculate the nonradiative decay rate of auramine - o and thioflavin - t , corresponding to a sphere of radius 2@xmath56 .
this happens to be just under half of the experimentally measured transition dipole moment for michler s hydrol blue@xcite .
the agreement in estimates of the rotor size is satisfying , since one is estimated from a hydrodynamic model of the non - radiative decay time@xcite , while the other is obtained from an absorbance spectroscopy experiment@xcite .
we will now summarize the ideas that have gone into the model hamiltonian equation [ eq : ham_mat ] .
* the model is built on a hilbert space spanned by structures with distinct charge and @xmath12-bond localization .
this notion of the low - energy electronic structure of cyanines has a history in chemical theory that goes back to the early part of the last century@xcite@xcite .
* we have introduced the idea that @xmath12-bonds have barriers to rotation that are not present in single ( @xmath57 ) bonds .
this concept is not new , and is one of the most recognized basic facts of structural organic chemistry .
the functional form of the barrier to rotation that we have used can be derived from valence - bond theoretical arguments@xcite .
the @xmath12-bond stabilization energy can be estimated to be in the range 1 - 2ev based on textbook arguments@xcite . *
the inter - ring coupling is a product of transfers onto and off of the bridge .
this reflects the idea that the direct ( i.e. single - particle ) inter - ring charge transfer coupling is negligible , which goes back as far as pauling@xcite .
since the direct coupling vanishes , the coupling goes as a product of transfers across the bridge bonds in lowest order .
the dependence on the variables @xmath15 and @xmath16 across each bond go as the overlap of local @xmath12 fragment orbitals across the respective bond . a lucid early description of the many - particle origin of the large ( @xmath58ev ) coupling in cyanines
was given by moffitt@xcite .
essential aspects of this analysis have recently been confirmed again using a quasi - diabatic analysis of the electronic structure of series of monomethine cyanines@xcite .
* we have introduced the notion that the solvent has different equilibrium positions for the different diabatic states .
the solvation interaction is described using a model of charge transfer between conducting spheres in a linear dielectric , with sphere radius and inter - sphere distances characteristic of the monomethine cyanine .
this model , which is quite well - established in the chemical physics literature related to electron - transfer reactions@xcite , has also been used in a description of anomalous solvatochromism in charged push - pull polyenes ( asymmetric polymethine cyanines)@xcite .
we have neglected the contribution of solvation effects to the coupling element , as has also been done in other studies of nonequilibrium solvation during photoisomerization reactions@xcite .
the potential energy surfaces given by the model can be qualitatively classified according to the `` phase diagram '' in figure [ fig : phase_diagram ] .
the figure classifies potential energy surfaces according to three yes / no questions .
these are : * will tict states form spontanously after electronic excitation ? * is there a barrier to twisting on the excited state surface ?
* do gapped tict states exist ?
a tict state is gapped if it can not be brought into coincidence with a conical intersection seam for any configuration of the solvent .
for parameters corresponding to typical monomethine cyanines in solvents of weak to moderate polarity , our model describes favorable ( spontaneous ) formation of twisted internal charge - transfer ( tict ) states on the excited - state pes .
twisting should be barrierless for dyes with electronic excitation energy above the far - red , but a barrier can exist for monomethines with weak electronic coupling or with weak @xmath12-bonding interactions .
numerical estimates that we have performed for limiting cases suggest the barrier , if it exists , is probably small ( a few @xmath59 , thermal energy at room temperature ) .
our model can describe the differential stabilization of twisting channels by the solvation field in a monomethine , and can also describe movement of the conical intersection between the twisting channels in response to changes in solvent configuration .
this is an important aspect of the model , which is worth highlighting now .
this behavior is summarized in figure [ fig : movingci ] .
this shows that our model , despite its simplicity , can describe the environmental selection of different twisting channels by an environmental field , such as has drawn interest in the literature on fluorescent proteins@xcite and photoactive yellow proteins@xcite .
our model suggests that , for typical monomethine cyanines with modest - sized endgroups ( a single substituted heterocycle ) in typical solvents ( for which we expect @xmath60 ) , there is no minimum on the excited state potential energy surface .
the bottom of the excited state potential corresponds to a conical intersection seam .
this seam grows along the `` valley floor '' of the excited state pes . for monomethine cyanines with modest - sized rings ( containing one substituted ring , @xmath61 ) , there are no gapped tict states in solvents of even moderate polarity . for monomethine cyanines in weak solvents and/or with large rings ,
our model predicts that gapped tict states may exist .
if diffusion within the channels is slow enough , then these gapped tict states may transiently accumulate excited - state population .
the twisted internal charge - transfer configurations / states in our model occur where the coupling matrix element vanishes @xmath62 .
this occurs on the lines @xmath63 and @xmath64 , limiting focus to the range indicated in [ fig : atlas ] .
we will divide the tict states by dividing them into two `` channels '' : the r channel ( @xmath64 ) and the l channel ( @xmath63 ) .
the points corresponding to these channels are also indicated in figure [ fig : atlas ] .
tict states will only form spontaneously on a relevant timescale if they are energetically favorable . for a single coordinate
, this corresponds to either of the two left panels of figure [ fig : tictoutcomes ] .
the excited - state energy ( relative to * fc * ) @xmath65 is defined in the @xmath49 and @xmath66 channels as @xmath67 the absolute value function introduces a derivative discontinuity where it s argument is zero . in this case , the argument is the adiabatic gap , and the derivative discontinuity gives the position of a conical intersection within the tict channels .
the excited - state energy in the @xmath66 channel can be written analytically on either side of the conical intersection as @xmath68 and in the @xmath49 channel as @xmath69 .
the situation is described visually in figure [ fig : rchannel ] , which shows the energies in the @xmath49 channel for some solvent configuration that puts the ci in that channel .
since the coupling vanishes in the tict channels , the diabatic crossing point is also a conical intersection between adiabatic states .
we can see immediately that diabatic state @xmath70 favors unwinding the angle @xmath15 , while @xmath71 is invariant to @xmath15 .
the coupling vanishes in the channel , so the lowest adiabatic excited state energy in the channel can never be lower than the adiabatic excited state energy at the geometry * rt * , corresponding to @xmath72 .
it follows that the lowest excited - state energy ( relative to * fc * ) in the @xmath49 channel equals the excited - state energy at * rt*. analogous reasoning says that the lowest energy in the @xmath66 channel is the energy at * lt*. the energies at these points is written on either side of the conical intersection as @xmath73 , while the excited - state energy at * rt * , which similarly bounds the energy in the @xmath49 channel from below , we have @xmath74 .
we can see that the lowest energy at * rt * is achieved for @xmath75 at @xmath41 , whereas for @xmath76 it is achieved at @xmath77 , which is at the conical intersection .
it follows that the lowest energy that is achievable in the tict channels is the energy at * rt * ( or * lt * ) _ at the solvent configuration which brings the conical intersection to that point_. this energy is given ( relative to * fc * ) by @xmath78 .
this equation gives the lowest ( most favorable ) excited - state energy that can be obtained for any tict state in our model ( relative to * fc * ) . if it is negative , then there
exist tict states that are energetically accessible from * fc*. if it is positive , then the tict states are not accessible from * fc*. in the former case , the tict states are expected to form spontaneously in the excited state , on a timescale that will depend on the intervening potential ( cf .
figure [ fig : tictoutcomes ] , difference between two left - hand panels and next section ) .
in the latter case , the molecule will remain on the excited state in a region of strong radiative coupling and would be expected to decay by fluorescence . if the energy at * rt*/*lt * vanishes , then there is no driving force for twisting . in the case of a flat , barrierless surface , equation
[ eq : trot ] is still useful if the average torque over a radian is taken as thermal energy at room temperature , @xmath59 . in that case , for a rotor with @xmath79 in a medium with viscosity 1 cp , the resulting timescale is still in the picosecond range .
non - radiative decay should still win over fluorescence , even if the potential is completely flat ( for typical solutions at room temperature ) .
the excited - state twisting process in monomethine cyanines has been invoked as an example of a barrierless viscosity - controlled process.@xcite quantum chemistry calculations on monomethine cyanines and similar systems usually indicate no barrier or small barrier.@xcite@xcite@xcite the accepted limits of quantum chemistry estimates of excited - state energy differences are of the order of a few @xmath80 ( even where the born - oppenheimer approximation is appropriate ) , so predictions of a barrier smaller than this are not conclusive.@xcite here , we examine what can be expected , with respect to the presence or absence of a barrier to twisting , on the basis of the few enumerable assumptions included in our model .
the question of whether or not there is a barrier to twisting boils down to the sign of the eigenvalues of the second - derivative matrix ( hessian ) with respect to the twisting displacements at * fc*. the solvent coordinate is always bound in the adiabatic excited - state of our model ( as can be verified by consideration of figure [ fig : nitzan ] and remembering that @xmath81 ) .
the hessian is diagonal at * fc * , and the force constants associated with the two twisting displacements are equal by symmetry .
the excited - state force constants with respect to these displacements have the particularly simple form @xmath82 .
negativity of the twisting force constants at * fc * implies accessibility of tict states , as can be seen by comparing equations [ eq : dble_derivativ ] and [ eq : e1min ] . if @xmath83 , then the model predicts barrierless descent to the tict manifold .
this would correspond to the case of the far left panel in figure [ fig : tictoutcomes ] .
if @xmath84 , then * fc * is bound but the tict states are still favourable , and twisting proceeds with a barrier ( middle panel of figure [ fig : tictoutcomes ] ) . if the tict states are inaccessible ( @xmath85 ) , then the molecule will decay by fluorescence ( right panel , figure [ fig : tictoutcomes ] ) . since the adiabatic transition is dipole allowed at * fc * , but forbidden in the tict channels , the transient fluorescence decay measures exit from the region of * fc*.
we expect the presence of the barrier to lengthen the fluorescence decay . however , the time to internal conversion on the far side of the barrier should be independent of the barrier height , so that in the activated regime we expect other contributions to the ground - state recovery time to be unaffected . the ground - state recovery time
would be expected to be dominated by the fluorescence decay time in this case .
comparison of equations [ eq : e1min ] and [ eq : dble_derivativ ] shows that @xmath86 implies that twisting is unfavorable in sufficiently non - polar solvents .
however , table [ tab : solv_reorg_en ] shows also that even for relatively non - polar solvents we still have @xmath87 . for michler s hydrol blue ( cf .
figure [ fig : scheme ] ) , we have the interesting case of @xmath88 , so that the driving torque for twisting in gas phase is expected to vanish .
we have tested this assertion for michler s hydrol blue using multi - reference perturbation theory estimates of the excited - state energy at the ground state minimum ( a model of * fc * ) and at a twisted excited - state minimum ( a model of * rt * or * lt * ) and found that this is true to within the expected accuracy of the calculations . in this case ,
solvation in weakly polar solvents such as eg .
chlorobenzene will induce a bias of @xmath89 , which should be sufficient to quench fluorescence .
the reddest dyes with absorbance in the visible would have @xmath90 .
we have numerically calculated the barrier height for a dye with @xmath91,@xmath92 with dimensions comparable to michler s hydrol blue ( @xmath93 , @xmath94 ) in chlorobenzene ( @xmath95 ) and found it to be less than @xmath96 .
assuming an attempt frequency of @xmath97 for a torsional barrier , this would not be sufficient to engender steady - state fluorescence in the dye for moderately polar solvents @xmath98 .
the observation of a barrier to twisting in our model must be approached with awareness that we are leaving out potentially important intramolecular degrees of freedom .
for example , our model does not consider bond vibration displacements , to which the states in these systems are known to be coupled .
laage and hynes have suggested a form for the diabatic coupling that is formally identical to the solvation model we use here.@xcite it has been pointed out that the coupling of solvation and stretching displacements through the same electronic degree of freedom ( the diabatic gap ) , can amplify the effects of either.@xcite if a solvation coordinate were included in this way , with a reorganization energy of order @xmath99 , it would act to reduce the barrier height and increase the driving force for reaction .
this may lower the barrier to the point that it becomes irrelevant for typical monomethine cyanines .
we conclude that our model predicts barrierless or near - barrierless excited state twisting for typical solvated monomethine cyanines , but we are unable to rule out the possibility that a monomethine cyanine may exist , if the dye has near - ir absorbance and large @xmath12-bonding interactions , that would show steady - state fluorescence in sufficiently non - polar solvents .
we call at particular tict state ( geometry ) _ gapped _ if it can not be made to coincide with the conical intersection seam for any configuration of the solvent .
it has been suggested that the timescales observed in ultrafast spectroscopic experiments on monomethine cyanines may be interpreted on the basis of a partitioning of the dynamics between ( fast ) descent into a basin corresponding to a twisted configuration , followed by diffusion to the sink or conical intersection.@xcite this notion presupposes the existence of an adiabatic twisted minimum on the excited state surface , in which diffusion could occur . in this section ,
we address the question of whether a true ( i.e. gapped ) excited state minimum can exist in the tict manifold . to answer the question , we analyze the extent and location of conical intersections in the model .
the conditions for intersection of adiabatic states is that the diabatic gap and coupling vanish simultaneously .
this gives @xmath100 for the coupling , and @xmath101 for the gap .
equation [ eq : ci_conditions1 ] is only ( and always ) satisfied in the tict channels , so that conical intersections will only ever occur in the channels .
equation [ eq : ci_conditions2 ] can be used to derive the solvent configuration that brings the conical intersection into coincidence with any point in the tict channels .
this is , for the @xmath66 and @xmath49 channels , @xmath102 this shows that the ci will be in the @xmath49 channel for @xmath103 and the @xmath66 channel for @xmath104 .
again , we have concentrated solely on the `` patch '' shown in figure [ fig : atlas ] ; a more general statement would include periodic images of the ci .
it is clear that [ eq : ci_conditions2 ] can only be solved for @xmath105 if @xmath106 , since the range of @xmath18 is limited to the interval @xmath107 $ ] . for solvent configurations @xmath108 ,
no conical intersection exists ( cf .
figure [ fig : movingci ] , bottom row ) . for any set of model parameters @xmath24 , @xmath23 , @xmath25 such that @xmath109 , the range of motion of the ci into the channels is limited to angles @xmath110 .
accordingly , there is an adiabatic gap remaining at angles @xmath111 less than some minimum angle @xmath112 for which the ci may exist .
if tict states exist that are gapped for all solvent configurations , then these may accumulate population on the excited state surface , which will decay following diffusion to the vicinity of the ci seam .
if the adiabatic excited state pes does possess a minimum , and if this basin is sufficiently removed in energy and configuration space from the intersection seam such that transient steady - state population accumulates there , we would expect the ground state recovery time to increase without consummate lengthening of the fluroescence decay time . this is because the tict states are dark , so accumulated population in the tict channels will not contribute to the fluorescence .
table [ tab : solv_reorg_en ] says that , for a dye with dimensions similar to michler s hydrol blue ( @xmath113 , @xmath114 ) , the reorganization energy should be of order 1 ev even for weakly polar solvents .
accordingly , we expect that @xmath115 for these solvents .
this implies that the conical intersection can access most of the tict configurations in the channels .
however , even for an angular interval of @xmath116 radian , we expect that the diffusion time in a medium of viscosity @xmath117 cp might be @xmath118 , and so this should be detectable as a delay in the ground state recovery time that is not apparent in the fluorescence decay time . for dye - solvent systems with @xmath119
, the ci can be brought into coincidence with any point in the tict channels . since the tict channels represent the bottom of the excited - state potential ( c.f .
discussion relevant to blue line in figure [ fig : phase_diagram ] , above ) , this implies that a ci spans the bottom of the excited state potential of the model in this regime .
this regime is the one that we expect to be relevant to most spectroscopic experiments using monomethine cyanines in moderate- to strongly polar solvents ( such as eg .
small - chain alcohols )
. the growth of access to the tict channels by the ci can be dramatically visualized by taking the fast - solvent limit , where the solvent is allowed to equilibrate to the dye at every fixed dye configuration .
this limit yields two - dimensional potential energy surface such as shown in figure [ fig : zippingupci ] for michler s hydrol blue . in the fast - solvent limit ,
the coincidence of the ci with the bottom of the excited - state pes is apparent as a `` zipping up '' of the tict states as the seam extends into the channels . the results shown in figure [ fig : zippingupci ] show that even for weakly polar solvents , the intersection can already extends over most of the tict configurations .
the pes given for acetonitrile in the figure is the same for _ all _ solvents with @xmath120 .
this is because solvent configurations that are sufficiently polarized to re - introduce an excited - state gap also raise the excited - state energy , so that the force on the solvent coordinate at these configurations will drive the system back towards the intersection .
accordingly , these solvent configurations are never sampled in the fast - solvent limit .
the phenomenon of `` zipping up '' of intersection seams in the fast - solvent limit has also been described by burgardt and co - workers for a model of the photoisomerization of a protonated schiff base@xcite .
this phenomenon can be traced in both cases to the diagonal form of the solvation coupling used in the hamiltonian equation [ eq : ham_mat ] .
erez et al .
have showed for the molecular rotors thioflavin - t and auramine - o , that the nonradiative decay timescale is 1 - 2 orders of magnitude slower than the transverse dielectric relaxation timescale in a variety of glass - forming solvents@xcite .
this is consistent with a twisting timescale that is of the same order as the _ longitudinal _ debye relaxation time .
the longitudinal time will an order of magnitude lower than the transverse time in lower alcohols .
the longitudinal relaxation time is the relevant time for solvation following a change in the charge distribution@xcite .
these points emphasize that only qualitative insights can be obtained from considering the fast - solvent limit , which is not believably achieved in real experiments on symmetric monomethines .
we have proposed a valence - bond model of non - equilibrium solvation effects on the excited - state twisting reaction of a general monomethine cyanine dye .
the model has been made intentionally simple , so as to explore the consequences of a few specific ideas .
the solvation model is similar to a previously published model of anomalous solvatochromism charged push - pull polyenes@xcite .
it differs by the introduction of potentials for twisting distinct @xmath12-bonds in the diabatic states , and the introduction of a twist - dependent coupling that goes as product of @xmath12 orbital overlaps across the bridge .
the functional forms used can be justified using valence - bond models@xcite .
our results suggest that the concept resonance in dyes , traditionally used to describe optical properties@xcite@xcite , may also explain non - radiative decay . since the resonance notation of figure [ fig : scheme ] describes a transfer between states with different charge- and bond - order localization , the location of the barrier is different in the diabatic states .
the correlation of charge and @xmath12-bond order localization has important consequences for the structure of the excited state pes in the model , and the response to non - equilibrium solvation .
the assumption of a coupling that varies as a product of overlaps across the bridge bonds is a consequence of the negligibility of the direct ( single particle ) inter - ring charge - transfer matrix element , which is traditionally assumed.@xcite the model broadly reproduces features of the ground and excited state pes that are observed in more complicated electronic structure calculations of monomethine cyanines@xcite@xcite and related systems@xcite@xcite@xcite@xcite@xcite .
these studies indicate that there are two energetically accessible twisting channels on the excited state surface , and that these channels are distinguished by the polarity of the adiabatic tict transitions within the channels@xcite@xcite , such that control of the pathway yield by environmental interactions is possible@xcite@xcite@xcite . for parameter regions characteristic of monomethine dyes ( c.f .
figure [ fig : scheme ] ) , there is a single charge - transfer conical intersection in the model that moves in response to changes in solvent configuration . for an unsolvated dye or a dye in an unpolarized solvent ,
the intersection occurs at a geometry of the dye where both bonds are twisted .
this is an ambiguous position with respect to the two tict channels .
it corresponds to either a concerted- or `` hula- '' twisted geometry@xcite . as the solvent configuration changes , so as to solvate better one or the other diabatic state , the intersection moves into one of the tict channels
this means that the intersection will occur at a geometry where one bond is twisted more than the other . in general , _ ab initio _
calculations on methine dyes show that minimal - energy conical intersections in symmetric monomethine cyanines occur at geometries where both bonds are twisted , but where the twist distribution is not symmetric@xcite .
this is also true in non - symmetric methines with very small diabatic gap@xcite .
the reason why our model predicts a symmetric twist distribution can be traced to the neglect of other intramolecular displacements that could break symmetry in the any real specific case . in any case
, our results suggest that intersections in methines can move in response to solvation , and will move to the bottom of the excited - state pes for solvents of moderate polarity .
decay via different twisting channels is favoured for distinct configurations of the solvent .
migration of conical intersections towards the bottom of the excited state potential in solvated organics was observed in early qm / mm wavepacket simulations of solvated photoisomerization in methines@xcite .
our model gives a simple and transparent description of this effect for the case of monomethine cyanines .
our model predicts that configurations with mixed twist distributions ( for example , the `` hula - twist '' configuration * ht * ) have relatively low energy on the excited - state pes .
in particular , these geometries are accessible from * fc * in gas phase .
this is contraindicated by _
ab initio _
calculations on monomethine cyanines@xcite , as well as non - symmetric monomethines with small diabatic gap@xcite .
the physical origin of the discrepancy is that , in our model , there is no additional penalty for breaking both @xmath12 bonds at once .
this could be added to our model easily , but would not qualitatively alter our main results .
we have provided a simple expression for the lowest energy possible in the tict channels , relative to * fc * , that can occur in our model .
this is equation [ eq : e1min ] .
this can be used in conjuction with the timescale formula equation [ eq : trot ] to yield a timescale for rotation of one of the rotors in a monomethine .
an estimate of the relevant @xmath11 in this case is the lowest energy in the channel ( equation [ eq : e1min ] ) divided by @xmath121 radians .
this yields the modified timescale formula @xmath122 . from this timescale , assuming no other non - radiative decay channels and also accounting for the existence of two rotors in monomethine cyanines , we can obtain an estimate of the fluorescence quantum yield @xmath123 as @xmath124 .
for michler s hydrol blue ( @xmath125 ev , @xmath126 ev , @xmath127 , @xmath128 ) in methanol ( @xmath129 cp ) , dichloromethane ( @xmath130 cp ) and chlorobenzene ( @xmath131 cp ) , this approach yields quantum yield estimates of @xmath132 , @xmath132 and @xmath133 , respectively , assuming a radiative timescale of @xmath134 ns .
these estimates are in reasonable agreement with measured values of @xmath132 , @xmath135 and @xmath136 , respectively.@xcite equations [ eq : qyield ] , [ eq : trot2 ] and [ eq : e1min ] together suggest that the fluorescence quantum yield of a monomethine cyanine is limited by the relative value of the driving torque in the excited state and the hydrodynamic friction . for a given dye chromophore ( i.e. given @xmath23 , @xmath24 ) ,
the driving torque will be modulated by the reorganization energy as equation [ eq : e1min ] , so that decreasing the reorganization energy of the environment of the dye will , in the absence of other considerations , give greater fluorescence , as will increasing the viscosity .
alternatively , for a given medium , the driving force will be increased by reducing the electronic coupling or by increasing the hydrodynamic volume . assuming a @xmath12-bond stabilization energy of @xmath126 ev , as we have here , implies that only monomethines with near - ir absorbance will have inaccessible tict states .
polar solvation can only accelerate tict formation ( in one channel or another ) . in a structured environment , such as provided by a protein ,
it is possible that the elastic properties of the protein may provide an additional potential barrier , but there seems little scope for engendering fluorescence in fluid solution by the manipulation of the driving torque . on the other hand , if the rotation timescale is linear in the viscosity as suggested by equation [ eq : trot ] , and given that the timescale for diffusion on a flat potential of @xmath137 ps for e.g. michler s hydrol blue , one expects that a 1000-fold increase in the viscosity should be sufficient to achieve significant fluorescence .
this is consistent with changes in the fluorescence of michler s hydrol blue recorded in a series of water - glycerol mixtures sampling this viscosity range.@xcite it is also broadly consistent with the radiative lifetime of auramine - o and thioflavin - t over a broader range.@xcite both the driving torque and the hydrodynamic friction depend on the spatial extent of the endgroups .
the driving torque varies inversely with @xmath27 ( cf .
equation [ eq : reorg_e ] ) , while the hydrodynamic friction scales with the volume ( i.e. with @xmath138 ) .
accordingly , varying the spatial extent of the endgroups with bulky substituents may be an effective way to tune the solution fluorescence quantum yield of monomethine cyanines .
symmetric monomethines have been used as experimental models for the study of barrierless viscosity - controlled reactions@xcite . for the cyanine 1122c , yartsev and coworkers
have cited evidence of a thermalized excited state population@xcite .
specifically , a fast ( 1ps ) component in the dynamics has been attributed to twisting motion towards an excited state basin , followed by depletion from the basin through the sink on the longer ( 10ps ) timescale@xcite . for a dye such as 1122c , with rings about double that of michler s hydrol blue , the reorganization energy in alcohol solution may be sufficiently low as to allow the existence of gapped tict states , which could support adiabatic excited - state population diffusion . for a dye of this size in methanol
, we have estimated the timescale for tict formation ( as above , but with @xmath139 , @xmath140 ) , and found it to be several picoseconds .
however , the timescale for diffusion on a flat potential ( such as would be expected for population diffusing on the adiabatic excited state in a gapped region of the tict channels ) is several _ tens _ of ps .
for dyes in moderately polar solution , for which the driving torque will be @xmath141 , the timescale for tict formation may be distinctly faster than the timescale for flat potential diffusion of an adiabatically excited tict population .
this suggests that our model may describe the transient accumulation of adiabatically excited tict population for dyes with rings comparable to those in 1122c .
further elaboration of this possibility would seem to require time - dependent simulations , and is beyond the scope of this paper .
we have proposed and analysed a valence - bond model of non - equilibrium solvation in a monomethine cyanine dye . the lowest adiabatic transition in these systems
is traditionally described using a concept of resonance between quantum states with distinct charge and bond localization .
this results in strong coupling to the twist degrees of freedom and a twist - dependent charge - transfer character of the transition , all of which our model captures .
it also describes several features of the potential energy surfaces that have been observed with more complicated electronic structure models@xcite , including the occurrence of multiple low - energy tict channels on the excited state surface , correlation between twisting and charge - transfer polarization of the transition in the channels , and the existence of low - energy conical intersections in the tict channels .
we believe that our model may be the simplest that captures these three effects .
our modelling suggests that solvent effects are significant on the potential energy surfaces that govern photoexcited twisting in monomethine cyanines .
polar solvation generally stabilizes tict states , and can increase the driving forces for tict formation .
we have provided a simple expression for the lowest possible energy ( relative to * fc * ) that can exist in the tict channels of our model ( equation [ eq : e1min ] ) .
our modelling suggests that tict formation will be favorable for typical monomethine cyanines in solvents sufficiently polar to allow good solvation .
tict formation should be barrierless for dyes with absorbance above 600 nm ( for which @xmath142 ev ) . for dyes with absorbance at lower photon energy ,
our modelling predicts a small barrier may exist on the excited - state surface .
this barrier should be less than a few @xmath143 for dyes with visible absorbance .
if a barrier exists , then our model also predicts that excited - state twisting will be unfavorable for the same dye in vacuum .
a conical intersection grows along the bottom of the excited - state surface in solvents with increasing relative dielectric . for monomethine cyanines with modest - sized rings in
moderately polar solvents , our modelling predicts that no adiabatically excited tict state can exist , as the bottom of the excited - state pes is completely spanned by a conical intersection seam
. rose , r. e. _ j. chem .
educ . _ * 1926 * , _ 3 _ , 973 brooker , l. _ rev . mod . phys . _ * 1942 * , _ 14 _ , 275 platt , j. r. _ j. chem .
* 1956 * , _ 25 _ , 80 platt , j. r. _ j. chem .
phys . _ * 1961 * , _ 34 _ , 862 kitts , c. ; beke - somfai , t. ; norden , b. _ biochemistry _ * 2011 * , _ 50 _ , 3451 lu , d. ; chen , g. ; perry , j. ; goddard iii , w. _ j. am .
soc . _ * 1994 * , _ 116 _ , 10679 thompson , w. ; blanchard - desce , m. ; hynes , j. _ j. phys .
* 1998 * , _ 102 _ , 7712 barzoukas , m. ; runser , c. ; fort , a. ; blanchard - desce , m. _ chem . phys .
lett . _ * 1996 * , _ 257 _ , 531 painelli , a. _ chem . phys . _ * 1999 * , _ 245 _ , 185 boldrini , b. ; cavalli , e. ; painelli , a. ; terenziani , f. _ j. phys . chem . a _
* 2002 * , _ 106 _ , 6286 cave , r. j. ; newton , m. d. _ chem .
lett . _ * 1996 * , _ 249 _ , 15 gorman , c. ; marder , s. _ proc .
u.s.a . _ * 1993 * , _ 90 _ , 11297 benson , r. c. ; kues , h. a. _ j. chem . eng . data _ * 1977 * , _ 22 _ , 379 hutten , e. h. ; pringsheim , p. _ j. chem .
phys . _ * 1945 * , _ 13 _ , 121 baldridge , a. ; feng , s. ; chang , y .- t . ; tolbert , l. m. _ acs combinatorial science _ * 2011 * , _ 13 _ , 214 baldridge , a. ; amador , a. ; tolbert , l. _ langmuir _ * 2011 * , _ 27 _ , 3271 paige , j. s. ; wu , k. y. ; jaffrey , s. r. _ science _ * 2011 * , _ 333 _ , 642 brey , l. ; schuster , g. ; drickamer , h. _ j. phys . chem . _ * 1977 * , _ 67 _ , 2648 clark , f. ; drickamer , h. _ j. phys . chem . _ * 1986 * , _ 90 _ , 589 murphy , s. ; sauerwein , b. ; drickamer , h. ; schuster , g. _ j. phys .
chem . _ * 1994 * ,
_ 98 _ , 13476 rettig , w. _ angew . chem . int .
edit . _ * 1986 * , _ 25 _ , 971 olsen , s. ; mckenzie , r. h. _ j. chem . phys . _ * 2012 * , _ 137 _ , 164319 amdursky , n. ; erez , y. ; huppert , d. _ acc . chem . res . _ * 2012 * , _ 45 _ , 1548 haidekker , m. a. ; theodorakis , e. a. _ j. biol
. eng . _ * 2010 * , _ 4 _ , 11 erez , y. ; amdursky , n. ; gepshtein , r. ; huppert , d. _ j.phys .
* 2012 * , _ 116 _ , 12056 van der meer , m. ; zhang , h. ; glasbeek , m. _ j. chem .
* 2000 * , _ 112 _ , 2878 erez , y. ; simkovitch , r. ; akulov , k. ; gepshtein , r. ; schwartz , t. ; huppert , d. _ j.phys .
c _ * 2014 * , _ 118 _ , 27063 ben - amotz , d. ; harris , c. b. _ j. chem .
phys . _ * 1987 * , _ 86 _ , 4856 ben - amotz , d. ; harris , c. b. _ j. chem . phys . _ * 1987 * , _ 86 _ , 5433 ben - amotz , d. ; jeanloz , r. ; harris , c. _ j. chem .
phys . _ * 1987 * , _ 86 _ , 6119 nitzan , a. _ chemical dynamics in condensed phases : relaxation , transfer , and reactions in condensed molecular systems _ ; oxford university press , 2006 ; pp 536549 barthel , j. ; bachhuber , k. ; buchner , r. ; hetzenauer , h. _ chem .
lett . _ * 1990 * , _ 165 _ , 369 heisler , i.a . ; kondo , m. ; meech , s.r .
_ j. phys .
b _ * 2009 * , _ 113 _ , 1623 kondo , m. ; heisler , i.a . ; meech , s.r .
_ j. phys .
b _ * 2010 * , _ 114 _ , 12859 improta , r. ; santoro , f. _ j. chem .
theory comput . _
* 2005 * , _ 1 _ , 215 dekhtyar , m. ; rettig , w. ; rozenbaum , v. _ j. photochem .
* 1999 * , _ 120 _ , 75 olsen , s. ; mckenzie , r. h. _ j. chem .
phys . _ * 2009 * , _ 131 _ , 234306 sanchez - galvez , a. ; hunt , p. ; robb , m. a. ; olivucci , m. ; vreven , t. ; schlegel , h. b. _ j. am .
_ * 2000 * , _ 122 _ , 2911 olsen , s. _ j. chem .
comput . _ * 2010 * , _ 6 _ , 1089 gromov , e. v. ; burghardt , i. ; koppel , h. ; cederbaum , l. s. _ j. phys . chem .
* 2011 * , _ 115 _ , 9237 ko , c. ; virshup , a. m. ; martinez , t. j. _ chem .
lett . _ * 2008 * , _ 460 _ , 272 knoch , f. ; morozov , d. ; boggio - pasqua , m. ; groenhof , g. _ comput . theor . chem .
_ * 2014 * , _ 1040 - 1041 _ , 120 bernardi , f. ; olivucci , m. ; robb , m. _ chemical society reviews _ * 1996 * , _ 25 _ , 321 klessinger , m. _ angew . chem .
_ * 1995 * , _ 34 _ , 549 levine , b. ; martinez , t. _ ann .
chem . _ * 2007 * , _ 58 _ , 1 matsika , s. ; krause , p. _ annu .
chem . _ * 2011 * , _ 62 _ , 621 schapiro , i. ; ryazantsev , m. ; ding , w. ; huntress , m. ; melaccio , f. ; andruniow , t. ; olivucci , m. _ aust . j. chem . _ * 2010 * , _ 63 _ , 413 haas , y. ; zilberg , s. _ j. photochem . photobiol . a _ * 2001 * , _ 144 _ , 221 virshup , a. m. ; chen , j. ; martinez , t. j. _ j. chem .
* 2012 * , _ 137 _ , 22a519 coe , j. d. ; ong , m. t. ; levine , b. g. ; martinez , t. j. _ j. phys . chem .
* 2008 * , _ 112 _ , 12559 toniolo , a. ; granucci , g. ; martinez , t. _ j. phys . chem . a _ * 2003 * , _ 107 _ , 3822 virshup , a. ; punwong , c. ; pogorelov , t. ; lindquist , b. ; ko , c. ; martnez , t. _ j. phys . chem .
b _ * 2009 * , _ 113 _ , 3280 burghardt , i. ; cederbaum , l. ; hynes , j. t. _ faraday disc . _ * 2004 * , _ 127 _ , 395 burghardt , i. ; hynes , j. t. _ j. phys . chem . a _ * 2006 * , _ 110 _ , 11411 spezia , r. ; burghardt , i. ; hynes , j. t. _ mol .
_ * 2006 * , _ 104 _ , 903 malhado , j. ; spezia , r. ; hynes , j. t. _ j. phys .
a. _ * 2011 * , _ 115 _ , 3720 fonseca , t. ; kim , h. j. ; hynes , j. t. _ j. mol .
liquids _ * 1994 * , _ 60 _ , 161 fonseca , t. ; kim , h. j. ; hynes , j. t. _ j. photochem .
* 1994 * , _ 82 _ , 67 kim , h. j. ; hynes , j. t. _ j. photochem . photobiol .
* 1997 * , _ 105 _ , 337 nordio , p. l. ; polimeno , a. ; saielli , g. _ j. photochem . photobio . _
* 105 * , 269 sissa , c. ; jahani , p. m. ; soos , z. g. ; painelli , a. _ chemphyschem _ * 2012 * , _ 13 _ , 2795 campo , j. ; painelli , a. ; terenziani , f. ; van regemorter , t. ; beljonne , d. ; goovaerts , e. ; wenseleers , w. _ j. am . chem .
soc . _ * 2010 * , _ 132 _ , 16467 thompson , w. ; blanchard - desce , m. ; alain , v. ; muller , j. ; fort , a. ; barzoukas , m. ; hynes , j. _ j. phys .
* 1999 * , _ 103 _ , 37663771 erez , y. ; liu , y .- h . ;
amdursky , n. ; huppert , d. _ j. phys .
* 2011 * , _ 115 _ , 84798487 laage , d. ; thompson , w. h. ; blanchard - desce , m. ; hynes , j. t. _ j. phys .
* 2003 * , _ 107 _ , 6032 tennyson , j. ; murrell , j. _ nouv . j. chim . _ * 1981 * , _ 5 _ , 361 fox , m. a. ; matsen , f. a. _ j. chem . educ . _
* 1985 * , _ 62 _ , 367 bernardi , f. ; olivucci , m. ; mcdouall , j. j. w. ; robb , m. a. _ j. chem . phys . _ * 1988 * , _ 89 _ , 6365 shaik , s. ; hiberty , p. c. in _ reviews in computational chemistry _ ; lipkowitz , k. b. , larter , r. , cundari , t. r. , eds . ;
john wiley & sons , inc .
, 2004 ; chapter 1 , pp 1100 liu , r. s. h. ; browne , d. t. _ acc .
res . _ * 1986 * , _ 19 _ , 42 griffiths , j. _ colour and constitution of organic molecules _ ; academic press professional , inc . ,
pp 240270 bortolus , p. ; cauzzo , g. _ trans .
soc . _ * 1970 * , _ 66 _ , 1161 cotton , f. a. _ chemical applications of group theory _ ; john wiley & sons , 1990 ; pp 438440 laage , d. ; burghardt , i. ; sommerfeld , t. ; hynes , j. t. _ chem .
chem . _ * 2003 * , _ 4 _ , 61 nitzan , a. _ chemical dynamics in condensed phases : relaxation , transfer , and reactions in condensed molecular systems _ ; oxford university press , 2006 ; pp 552602 rettig , w. _ j. mol .
struct . _ * 1982 * , _ 84 _ , 303 akesson , e. ; bergstrm , h. ; sundstrm , v. ; gillbro , t. _ chem .
lett . _ * 1986 * , _ 126 _ , 385 looney , c. ; simpson , w. _ j. am .
* 1954 * , _ 76 _ , 6293 amdursky , n. ; gepshtein , r. ; erez , y. ; huppert , d. _ j. phys .
* 2011 * , _ 115 _ , 2540 pauling , l. _ proc . natl .
* 1939 * , _ 25 _ , 577 moffitt , w. _ proc . phys .
* 1950 * , _ 63 _ , 700berg , u. ; kesson , e. ; alvarez , j. l. ; fedchenia , i. ; sundstrm , v. _ chem . phys . _ * 1994 * , _ 183 _ , 269 olsen , s. ; smith , s. _ j. am .
soc . _ * 2008 * , _ 130 _ , 8677 weigel , a. ; pfaffe , m. ; sajadi , m. ; mahrwald , r. ; improta , r. ; barone , v. ; polli , d. ; cerullo , g. ; ernsting , n. p. ; santoro , f. _ phys .
phys . _ * 2012 * , _ 14 _ , 13350 azizi , z. ; roos , b. r. o. ; veryazov , v. _ phys .
_ * 2006 * , _ 8 _ , 2727 dietzek , b. ; tarnovsky , a. n. ; yartsev , a. _ chem .
phys . _ * 2009 * , _ 357 _ , 54 boggio - pasqua , m. ; robb , m. ; groenhof , g. _ j. am . chem
. soc . _ * 2009 * , _ 131 _ , 13580 toniolo , a. ; olsen , s. ; manohar , l. ; martnez , t. _ faraday discuss . _ * 2004 * , _ 127 _ , 149 dietzek , b. ; yartsev , a. ; tarnovsky , a. n. _ j. phys . chem _ * 2007 * , _ 111 _ , 4520 dietzek , b. ; christensson , n. ; pascher , t. ; pullerits , t. ; yartsev , a. _ j. phys .
b. _ * 2007 * , _ 111 _ , 5396 momicchioli , f. ; baraldi , i. ; carnevali , a. ; caselli , m. ; ponterini , g. _ coord .
_ * 1993 * , _ 125 _ , 301 some symmetric monomethine cyanine dyes .
the ground state of these cyanines is an even superposition of states @xmath144 and @xmath145 with different charge and bond localization . in michler s hydrol blue ( upper left ) the physical charge carrier is an electron , but a hole in other dyes shown , explaining the different disposition of bond and formal charge.,width=377 ] three distinct possible outcomes for an excited - state twisting reaction leading to a dark twisted internal charge - transfer ( tict ) state .
the reaction coordinate is taken to be a single bond - twisting angle @xmath146 , which connects the franck - condon ( fc ) region on the excited - state potential energy surface with the tict state at @xmath147 . _
( left ) _ if the reaction is downhill and barrierless , fast non - radiative decay is expected , with fast ( fs - ps ) decay of fluorescence in solutions of low ( 1 cp ) viscosity . _
( middle ) _ if there is a barrier along the reaction coordinate , but the reaction is still favorable , then slower non - radiative decay is expected , with the rate depending on the barrier height . for barriers less than a few thermal energy units @xmath148
, only transient fluorescence is expected ; for larger barriers steady - state fluorescence may be observed .
_ ( right ) _ for an uphill reaction , the tict states do not form spontaneously in significant amounts . in this case , the dye will decay by emission of fluorescence.,width=491 ] a map of the `` asymmetric unit '' in dye configuration space , with labelling of landmarks referred to in the text .
the space , spanned by the twisting angles shown at top , is periodic and infinite . in this paper
, we focus solely on the patch shown , which is spanned by intervals of @xmath149 $ ] in both the twisting angles @xmath15 and @xmath16 .
the corners of the patch are landmarks frequently mentioned by acronymn in the text : the franck - condon ( * fc * ) configuration , the left - twisted configuration ( * lt * ) , the right - twisted configuration ( * rt * ) and a `` hula - twist '' ( * ht * ) configuration@xcite where both bonds are twisted
. initial photoexcited population would be prepared at * fc * because the transition is bright there and it coincides with the ground state minimum .
the boundaries of the patch opposing * fc * and connecting the twisted configurations are where the twisted internal charge - transfer ( tict ) configurations are found .
the tict configurations are divided into the `` l '' tict channel , consisting of configurations where @xmath63 , and the `` r '' tict channel @xmath150 .
conical intersections between the adiabatic states in our model only occur in the tict channels
. our model does nt distinguish concerted ( even ) and `` hula '' ( odd ) combinations of the torsions , nor does it distinguish between different _
cis_/_trans _ isomers of a given dye.,width=264 ] a schematic diagram of the solvation interaction in our model . shown
are the ( quadratic ) potential energy surfaces of the diabatic and adiabatic states along a single solvation coordinate _ s _ , which gauges the polarization of the solvent .
the solvent is in equilibrium with diabatic state @xmath70 ( green line ) at @xmath41 and in equilibrium with state @xmath71 ( red line ) at @xmath42 .
the diabatic gap is modulated by the torsion angles of the dye , @xmath15 and @xmath16 ( cf .
figure [ fig : atlas ] ) through the torsion coupling constant @xmath24 , which is the energy of twisting the @xmath151 to the bridge .
the splitting of the adiabatic ground ( blue ) and excited ( orange ) states has contributions from the diabatic gap and the diabatic coupling matrix element .
the diabatic coupling is modulated by the torsion angles through the electronic coupling constant @xmath23 , and is equal to half of the electronic excitation energy at the ground state ( planar ) minimum . at the solvent configuration where the diabatic states cross ,
the adiabatic states are split by an amount equal to twice the diabatic coupling .
the diabatic solvent reorganization energy @xmath25 is the energy required to displace one diabatic state from its own equilibrium solvent configuration to the equilibrium configuration of the other .
it is the same for both diabatic states .
, width=377 ] this `` phase diagram '' summarizes the qualitative features of potential energy surfaces ( pess ) provided by the model in distinct regimes of the parameters @xmath152 .
the parameter regimes characteristic of typical monomethine cyanines are inside the dotted ellipse .
pes given by the model are classified according to the disposition with respect to three lines drawn . above the blue line , spontanous formation of dark tict states
is expected . below the line
, the dye will decay by fluorescence .
to the left of the green line , the ( planar ) * fc * configuration is stable ( left ) on the exited - state pes .
if tict states are still favorable , then there is a barrier to reach them . to the right of the line the planar ( fc )
state is unstable on the excited state surface , and twisting occurs without barrier . below the red line , there are tict states that are adiabatically gapped for all solvent configurations . above the red line
, the tict states are ungapped , in the sense that the ci can be made to coincide with any tict geometry for some configuration of the solvent.,width=453 ] changes in the potential surfaces of the dye in response to changes in the solvent configuration . for an unpolarized solvent , such as at the ground state minimum ,
the excited state pes is ambiguous with respect to the l and r tict channels ( c.f .
[ fig : atlas ] ) , and the conical intersection ( ci ) between adiabatic states is positioned ambiguously , where both bonds are equally twisted . as the solvent polarizes in one or the other direction ,
one channel is stabilized over another and the ci moves into the stabilized channel .
this corresponds to positioning the ci at dye configurations where one bond is twisted ( @xmath147 ) and the other is less twisted ( @xmath153 ) . for moderate polarization of the solvent ,
the ci is located in one of the twisting channels . at a critical polarization @xmath154 , the ci reaches the `` end '' of the tict channel , where one bond is twisted and the other planar . for polarizations beyond this , the adiabatic gap reappears and the excited state
is destabilized .
the solvent coordinate is bound in the interval @xmath107 $ ] , so that not all regimes are accessible if @xmath75 . the parameters used here were chosen for michler s hydrol blue ( @xmath155 ) in acetonitrile ( @xmath156).,width=453 ] movement of the conical intersection in the tict channels can be understood by examining the diabatic states within a single channel ( here the r channel , for which @xmath64 , for a solvent configuration favoring @xmath70 ) . _ the diabatic coupling vanishes everywhere in the figure _ , so the adiabatic states are given by taking the lowest and highest value at every point , and the crossing point is a conical intersection between adiabatic states . the potential for diabatic state @xmath70 in the channel is flat , because the @xmath12-bond in this state is already twisted .
the potential for @xmath71 rises as @xmath157 . for an unpolarized solvent ( such as at the ground state minimum ) , the ci is located at ht , a configuration where both bonds are twisted . as the solvent moves to stabilize @xmath71 , the excited state energy at lt decreases , and the ci moves from ht towards rt ( to the left of the figure ) .
when the solvent stabilization @xmath158 , the ci has reached rt . for larger polarizations of the solvent , the @xmath12-bond stabilization energy can no longer compete with the energy of the solvent interaction .
then , the adiabatic gap reopens and the ci disappears.,width=377 ] potential energy surfaces for michler s hydrol blue ( @xmath159 ev , @xmath160 ev ) in the fast - solvent limit , in gas phase ( left ) , chlorobenzene ( middle ) and acetonitrile ( right ) .
reorganization energies and static dielectric constants for each solvent are shown . in the fast solvent limit , growth of the ci seam along
the bottom of the excited state potential energy surface ( pes ) appears as a `` zipping up '' of the ci seam along the twisted internal charge - transfer ( tict ) states . this is because , for any tict configuration , the solvent configuration that yields the lowest excited - state energy is the one that brings the ci to that point .
the fast - solvent limit is taken at any dye configuration by letting the solvent equilibrate to the excited state of the dye at that configuration .
shown are fast - solvent - limit pes for michler s hydrol blue ( @xmath160 ev , @xmath161 ev , @xmath162 @xmath56 , @xmath163 @xmath56 ) in gas phase ( left ) , a weakly polar solvent ( chlorobenzene , middle ) and a strongly polar solvent ( acetonitrile , right ) .
even for chlorobenzene , the ci seam spans most of the tict configurations .
the pes shown for acetonitrile is identical in any solvent for which @xmath164 , since at that point the tict states are already `` all zipped up '' ( @xmath165 ev for this case ) .
the fast - solvent limit is not physically realistic for these systems , but provides good illustration of the growth of the ci seam in solvents with increasing susceptibility .
, width=566 ] | we propose and analyze a two - state valence - bond model of non - equilibrium solvation effects on the excited - state twisting reaction of monomethine cyanines .
suppression of this reaction is thought responsible for environment - dependent fluorescence yield enhancement in these dyes .
fluorescence is quenched because twisting is accompanied via the formation of dark twisted intramolecular charge - transfer ( tict ) states .
for monomethine cyanines , where the ground state is a superposition of structures with different bond and charge localization , there are two possible twisting pathways with different charge localization in the excited state . for parameters corresponding to symmetric monomethines ,
the model predicts two low - energy twisting channels on the excited - state surface that lead to a manifold of twisted intramolecular charge - transfer ( tict ) states . for typical monomethines , twisting on
the excited state surface will occur with a small barrier or no barrier .
changes in the solvation configuration can differentially stabilize tict states in channels corresponding to different bonds , and that the position of a conical intersection between adiabatic states moves in response to solvation to stabilize either one channel or the other .
there is a conical intersection seam that grows along the bottom of the excited - state potential with increasing solvent polarity . for monomethine cyanines with modest - sized terminal groups in moderately polar solution ,
the bottom of the excited - state potential surface is completely spanned by a conical intersection seam .
# 1 @xmath0 # 1 # 1 # 1 @xmath1 # 1 # 1 # 1#2 @xmath2 # 1#2 # 1 .
# 2 # 1#2#3 @xmath3 # 1#2#3 # 1 # 2 # 3 # 1#2#3 @xmath4 # 1#2#3 # 1 # 2 # 3 # 1#2#3#4#5#6 @xmath5 # 1#2#3#4#5#6 ( ) # 1#2#3#4#5#6 @xmath6 # 1#2#3#4#5#6 \ { } # 1#2#3#4#5#6#7#8#9 \ { } [ 1]d .. #1 |
a general hamiltonian system may have two different kinds of constants of motion ( first integrals ) ; usual ones are caused by symmetries of the hamiltonian , while the others , so - called _ casimir invariants _ , are independent of a specific choice of hamiltonian , but are pertinent to the underlying poisson algebra@xcite .
a casimir invariant @xmath0 is a function ( observable ) such that @xmath1 for every @xmath2 ( here @xmath3 denotes the poisson bracket of a hamiltonian system ) , i.e. , @xmath0 belongs to the center of the poisson algebra .
the existence of a nontrivial casimir invariant ( @xmath4 constant ) is the determining characteristics of a _ noncanonical _ hamiltonian system ( or , a degenerate poisson algebra ) .
conversely , a canonical hamiltonian system does not have a casimir invariant ; all constants of motion are of the first kind .
interestingly , we may often convert a noncanonical system into a canonical system .
then , what will become of casimir invariants ? here
we put the method of _ clebsch parameterization _ into the perspective .
the noncanonical hamiltonian system to be examined is that of the ideal magnetohydrodynamics ( mhd)@xcite . a hamiltonian formulation of fluid / plasma mechanics is generally noncanonical when it is described by eulerian variables , and has two kinds of casimir invariants@xcite ; one is the total mass , and the others are helicities . by representing the the vector fields ( fluid velocity and magnetic field ) in terms of clebsch parameters , we can rewrite the evolution equation as a canonical hamiltonian system , and provide it with an action principle @xcite .
the aim of this work is to elucidate how the original casimir invariants translate in the canonized system
. we will show that they become the first - kind constants of motion corresponding to the gauge symmetries of the clebsch parameterization@xcite .
the gauge transformation also conserves the action integral ; hence the constants of motion are _ noether charges _ ,
i.e. , the casimir invariants translate as the noether charges corresponding to the gauge symmetry of clebsch parameterization .
we organize this paper as follows : in section[sec : formulation ] , we start by reviewing the standard noncanonical hamiltonian formulation of mhd ( subsection[subsec : noncanonical ] ) and its canonization by clebsch parameterization ( subsection[subsec : canonical ] ) . in section[sec : symmetry ] , we elucidate the symmetry of the canonized system . the hamiltonian flow generated by
the original casimir invariant defines a symmetry group ( subsection[subsec : symmtry_group ] ) ; the symmetry turns out to be the gauge symmetry ( redundancy ) of the clebsch parameterization ( subsection[sebsec : gauge_group ] ) , which yields a noether charge that reproduces the original casimir invariant ( subsection[subsec : noether ] ) . in subsection[subsec : generalization ] , we derive an extended set of constants of motion by studying a generalized class of symmetries .
section [ sec : conclusion ] concludes this work with a remark on the lagrangian formalism and the relabeling symmetry .
the mhd system is described by @xmath5 where @xmath6 is the density , @xmath7 is the velocity , @xmath8 is the magnetic field , and @xmath9 is the specific enthalpy .
all variables are normalized by the alfvn unit , i.e. , the energy densities ( thermal @xmath9 , kinetic @xmath10 and magnetic @xmath11 ) are normalized by the representative magnetic energy density @xmath12 ( @xmath13 is the vacuum permeability ) . here
assume a barotropic relation @xmath14 .
we consider a smoothly bounded , simply connected domain @xmath15 .
we assume @xmath16 on the boundary @xmath17 ( @xmath18 is the normal unit vector onto @xmath17 ) .
we may cast the mhd equations ( [ mhd-1])-([mhd-3 ] ) in a hamilton form .
a general hamilton s equation is written as @xmath19 where @xmath20 ( phase space ) is a state vector , @xmath21 is a poisson operator , and @xmath22 is a hamiltonian .
the adjoint representation of the dynamics reads @xmath23 where @xmath24 is an observable , and @xmath25 is the poisson bracket induced by @xmath26 ( we denote by @xmath27 the inner product of the hilbert space @xmath28 ) . for mhd
, we define @xmath29 where @xmath30 the specific thermal energy , @xmath31 is specific enthalpy . evidently , substituting ( [ mhd - hamiltonian-1])-([mhd - hamiltonian-3 ] ) into ( [ hamilton - general ] ) reproduces the mhd equations ( [ mhd-1])-([mhd-3 ] ) .
there are three casimir invariants ( hence , the mhd system is noncanonical ) : @xmath32 where @xmath33 is the vector potential ( @xmath34 is the inverse operator of @xmath35 , which is a self - adjoint operator ; see@xcite ) ; @xmath36 is the total mass , @xmath37 is the magnetic helicity , and @xmath38 is the cross helicity .
we can canonize the mhd system by parameterizing the vector fields @xmath7 and @xmath8 in terms of clebsch parameters . here
we invoke a generalized form of clebsch parameterization such that @xmath39 when @xmath40 is chosen to be the dimension of base space ( here , @xmath41 ) , ( [ clebsch ] ) gives a complete representation of general @xmath42 by @xmath43 , @xmath44 and @xmath45 that satisfy appropriate boundary condition independently@xcite . from now on , the contraction rule on the indexes will be used to abbreviate notation .
the formulation of clebsch - parameterized mhd will be introduced @xcite . here
we put @xmath46 the index @xmath47 runs over @xmath48 . at @xmath17 , we assume the following boundary conditions for the clebsch parameters : @xmath49 where @xmath50s are constant numbers , which are consistent to the boundary conditions ( [ bc - v ] ) and ( [ bc - b ] ) on the physical variables . denoting @xmath51 we define @xmath52 hamilton s equation reads as a canonical system @xmath53 where @xmath54 and @xmath55 .
we have a lagrangian ( represented by the clebsch fields ) @xmath56 by which the action is written as ( denoting @xmath57 ) @xmath58 where @xmath59 is a space - time domain @xmath60 .
the invariants , @xmath36 , @xmath37 and @xmath38 are no longer `` casimir invariants '' in the canonized system , while they must be still constants of motion .
delineation of how they are conserved in the canonized formalism is the aim of the next section .
let @xmath61 denotes the canonical bracket . given a hamiltonian @xmath2 ,
a constant of motion is a functional @xmath0 such that @xmath62 an infinitesimal transformation ( flow ) generated by a virtual hamiltonian @xmath63 is given by ( with small @xmath64 ) @xmath65 if @xmath63 and the actual hamiltonian @xmath2 commute , i.e. , if @xmath63 is a constant of motion , @xmath66 defines a _ symmetry group _ : ( [ conservation law ] ) implies @xmath67 to simplify the notation , we will omit @xmath64 in writing perturbations @xmath68 .
the symmetry groups generated by the `` original casimir invariants '' are @xmath69 where @xmath70 . here
we show that the symmetry groups ( [ total_mass_to_symmetry])-([cross_helicity_to_symmetry ] ) , generated by the original casimir invariants , are gauge transformations of the clebsch parameterization , i.e. , the physical variables @xmath6 , @xmath7 and @xmath8 are unchanged by the action of the group . under the transformation ( [ total_mass_to_symmetry ] )
, it is evident that @xmath71 and @xmath72 , because @xmath6 and @xmath8 are independent of @xmath73 .
also @xmath74 , because @xmath7 includes @xmath73 in the form of its gradient . under ( [ magnetic_helicity_to_symmetry ] ) ,
@xmath71 and @xmath72 are obvious , because @xmath6 and @xmath8 do not depend on @xmath73 , @xmath75 and @xmath76 .
we also find that @xmath77 vanishes : @xmath78 under ( [ cross_helicity_to_symmetry ] ) , @xmath71 is evident . @xmath77 and
@xmath79 are calculated as follows : @xmath80 to show that @xmath36 , @xmath37 and @xmath38 are noether charges of the canonized system , we have yet to prove that the gauge transformations ( the symmetry groups ) conserve the action .
the lagrangian density @xmath81 consists of two parts ; the canonical 1-form and @xmath82 . because the transformations ( [ total_mass_to_symmetry])-([cross_helicity_to_symmetry ] ) are the gauge transformations , it is obvious that @xmath82 is unchanged . on the other hand ,
the canonical 1-form may be changed by a gauge transformation .
however , the change is an exact differential , thus the action integral is changed only by a constant .
in fact , an arbitrary hamiltonian flow @xmath83 ( generated by a virtual hamiltonian @xmath63 ) conserves the canonical 1-form .
let us denote @xmath84 .
for @xmath85 and @xmath86 , we observe @xmath87 and @xmath88 which implies that @xmath89 is the exact differential .
explicitly , the action of ( [ magnetic_helicity_to_symmetry ] ) yields @xmath90 thus @xmath91 is an exact differential .
similarly , by the action of ( [ cross_helicity_to_symmetry ] ) , we obtain @xmath92 where @xmath93 .
let us briefly review how a noether charge is derived by an action - invariant transformation .
we consider a first - order lagrangian such that @xmath94 , which yields the euler - lagrange equation @xmath95 .
suppose that @xmath96 has a symmetry : @xmath97 ( with some vector @xmath98 ) under a transformation @xmath99 .
for a solution @xmath100 of the euler - lagrange equation , we obtain @xmath101 for a perturbation @xmath102 of a symmetry group , we may write @xmath103 where @xmath104 is a noether current , and @xmath105 is the corresponding noether charge .
applying this procedure to the canonized mhd lagrangian and the gauge symmetries , we obtain @xmath106 the third noether charge ( [ symmetry_to_cross_helicity ] ) is equal to the cross helicity @xmath38 ; using the boundary condition of @xmath8 , @xmath107 a transformation conserving a lagrangian maps a solution of the equation of motion to another solution .
however , the converse is not necessarily true in fact , there is a gauge transformation of the clebsch parameterization which does not conserve the action .
let @xmath108 , @xmath109 , and @xmath110 .
these perturbations yield a gauge transformation : @xmath111 it is also obvious that @xmath8 and @xmath6 are unchanged
. however , we obtain @xmath112 which is not an exact differential . we have elucidated a beautiful relation , mediated by the constants @xmath36 , @xmath37 and @xmath38 , between the eulerian noncanonical formalism and the clebsch - parameterized canonical formalism ; in the former , the constants are the casimir invariants that foliate the phase space , and in the latter , they are the noether charges that identify symmetry groups reflecting the redundancy of the clebsch parameterization ( see fig.[fig1 ] ) . ) and the casimir leaves ( level - sets of @xmath0 ) .
right : by the clebsch parameterization , the system is embedded in a larger phase space with redundant degree of freedom ; the hamiltonian group action generated by the noether charges delineates the symmetry built in the clebsch parameterization.,height=283 ] the clebsch parameterization enables us to discribe local conservation of mass and helicity .
we define two sets of functions @xmath113 and @xmath114 such that @xmath115 from ( [ equation_of_motion ] ) , @xmath116 and @xmath117 .
we easily verify @xmath118 when @xmath119 , @xmath120 is a constant of motion . from ( [ rule3 ] ) , @xmath121 and using ( [ rule2 ] ) and ( [ rule4 ] ) , @xmath122 ; hence we find a new constant of motion @xmath123 ( @xmath124 is an arbitrary smooth function ) . from ( [ rule4 ] ) and ( [ rule5]),@xmath125 . using ( [ rule1 ] ) , and summing up the terms such that @xmath126 and @xmath127
, we obtain @xmath128 . from ( [ rule4 ] ) , @xmath129 ,
thus @xmath130 is a constant of motion .
since @xmath131 is a scalar function that is constant along the streamlines of the flow @xmath7 , we may regard it as a weighting function co - moving with the plasma .
choosing @xmath132 to be a support function on a co - moving volume element , @xmath123 and @xmath130 measure local mass and helicity contained in the volume element .
these generalized constants of motion are related to the following gauge symmetries : @xmath133 where @xmath134 denotes the derivative of @xmath124 with respect to @xmath135 , and @xmath136 the one with respect to @xmath137 .
for example , @xmath77 under the transformation ( [ generalized_mass_and_symmetry_mhd ] ) is @xmath138 and @xmath71 and @xmath72 are obvious , thus ( [ generalized_mass_and_symmetry_mhd ] ) is a gauge transformation .
we have shown , in the context of mhd , that casimir invariants of a noncanonical hamiltonian system are converted , upon canonization by an alternative parameterization ( here , clebsch parameterization ) of the state vector , into noether charges pertinent to the gauge freedom ( redundancy ) of the new parameterization . while the conclusion delineates a simple relation between different formalisms of dynamical systems , the specific forms of the symmetries , embodying the concrete relation , are rather complicated .
since the clebsch parameterization implies a nonlinear variable transformation , the symmetries are not apparent ( excepting some simple ones ; cf .
@xcite ) . emphasizing another aspect of the problem , we may say that we could unearth , guided by the known casimir invariants , the general gauge symmetries of the clebsch parameterization .
there are some different methods of canonization .
a well - known method that applies for various fluid - mechanical systems is the usage of lagrangian variables to represent the dynamics of fluid elements ; the lagrangian variables label the initial position of each fluid element . formulating an action by a lagrangian of newtonian - mechanical continuum
, we obtain a canonical systems of hamilton s equation of motion@xcite ; see @xcite for the original formulation of the lagrangian action principle of mhd . in the canonized system of lagrangian variables , the casimir invariants are translated differently .
the total mass @xmath139 and the magnetic helicity @xmath140 are now trivially conserved , because they are the integrals of the _ attributes _ of fluid elements ; the local density and magnetic helicity are not dynamical variables , but are _ constants _ bound to each fluid element ( like charges of particles in particle systems ) . only the cross helicity @xmath141 is a nontrivial first integral ; it is now the noether charge pertaining to the _ relabeling symmetry _ that represents the arbitrariness of providing fluid elements with lagrangian labels@xcite .
since clebsch parameters may be viewed as the eulerian counterparts of lagrangian labels ( or , the pull - back along the cauchy characteristics of the fluid velocity @xmath7 ; see @xcite ) , it is natural that the gauge symmetry of the clebsch parameters ( present canonization ) and the relabeling symmetry of the lagrangian labels ( lagrangian canonization ) yield the same constant of motion .
the authors thank dr .
yohei kawazura for his suggestions and discussions .
this was supported by grant - in - aid for scientific research from the japanese ministry of education , science and culture no . | it is shown that the clebsch parameterization , upon canonizing the hamiltonian system of ideal fluid or plasma dynamics , converts the casimir invariants ( mass and helicities ) in the original noncanonical formulation into the noether charges pertinent to the gauge symmetries of the parameterization .
the problem is addressed in the context of magnetohydrodynamics .
the concrete forms of the gauge symmetries of clebsch parameterization are worked out . |
significant theoretical activity , aimed at understanding the properties of matter under extreme conditions , has been triggered recently by the heavy - ion collision experiments conducted at rhic and the lhc , in which the deconfined phase of qcd matter , the quark - gluon plasma , is created .
the transition from the hadronic to the deconfined , partonic phase is an analytic crossover at zero baryo - chemical potential @xmath0 @xcite with a transition temperature @xmath1 determined in lattice qcd simulations @xcite .
this crossover feature also extends to small values of @xmath0 .
the possibility that the transition becomes first - order for larger @xmath0 , which would imply the existence of a critical point , is currently investigated by the beam energy scan program at rhic , soon to be followed by the cbm experiment at fair .
event - by - event fluctuations of the net - electric charge and net - baryon number , which are conserved charges of qcd , are expected to become large near a critical point @xcite : for this reason , they have been proposed as ideal observables to verify its existence and to determine its position in the qcd phase diagram @xcite .
experimental results for these measures were recently reported for several collision energies @xcite .
in addition , as a consequence of the increasing precision achieved in the numerical simulations , it is becoming possible to extract the chemical freeze - out parameters ( i.e. freeze - out temperature @xmath2 and corresponding baryo - chemical potential @xmath3 ) from first principles , by comparing the measured fluctuation observables to corresponding susceptibility ratios calculated in lattice qcd @xcite . when compared to experimental data from heavy - ion collisions ,
present lattice simulations have , however , their limitations : they can not take the experimental acceptance cuts into account and they are available only for small chemical potentials . as a consequence of the latter , only the lowest order susceptibilities are available on the lattice at finite @xmath0 .
moreover , the experimental restriction of net - baryon to net - proton number measurements can not be realized on the lattice .
recently , fluctuation observables have been investigated in transport approaches @xcite as well as in various baseline studies within the hrg model @xcite .
calculations based on the hrg model in chemical equilibrium reproduce the equilibrium lattice qcd results for the susceptibilities and their ratios in the hadronic phase reasonably well @xcite . furthermore , the model allows to expand the range of @xmath0-values and consequently to calculate ratios of higher order susceptibilities at finite @xmath0 , as well as to implement kinematic acceptance cuts in rapidity @xmath4 , pseudo - rapidity @xmath5 and transverse momentum @xmath6 , thus , providing a valuable tool to extract the freeze - out conditions from the experimental data . in the past , statistical hadronization models ( shms ) have been used to analyze experimental data on particle production by comparing the data to thermal abundances calculated in hrg model approaches for all collision energies ranging from ags to the lhc , see e.g. @xcite and references therein .
the freeze - out temperatures determined in this way lie , however , at the upper limit of the uncertainty band of the lattice qcd results for @xmath1 @xcite , which is most pronounced for the highest collision energies . in this paper
, we calculate ratios of susceptibilities quantifying fluctuations in the number of net protons and in the net - electric charge within the hrg model : our study includes the effects of resonance decays and isospin randomization for the nucleons , as well as kinematic acceptance cuts in agreement with the experimental analysis .
we extract new freeze - out points in the ( @xmath7)-plane of qcd matter and compare them with the freeze - out curve determined in ref .
finally , we discuss the issue of the consistency between the freeze - out parameters obtained from the analysis of net - electric charge and net - proton fluctuations .
we find that it is possible to use a combined analysis of the lowest - order cumulants of the net - electric charge and the net - proton distributions in order to extract common freeze - out conditions @xmath2 and @xmath3 .
possible limitations of this method are addressed .
the hrg model provides a suitable description of the bulk properties of hadronic matter in thermal and chemical equilibrium , see e.g. @xcite . in the thermodynamic limit , the pressure as a function of temperature @xmath8 and all hadron chemical potentials @xmath9 is given by @xmath10 \
, , \label{equ : pressurehrg}\end{gathered}\ ] ] where the sum is taken over all hadronic states @xmath11 , including resonances , in the model ( baryons and anti - baryons being summed independently ) . in eq .
( [ equ : pressurehrg ] ) , @xmath12 and @xmath13 denote the degeneracy factor and the mass , respectively , and @xmath14 is the chemical potential of the species @xmath11 in chemical equilibrium .
@xmath15 , @xmath16 and @xmath17 are the respective quantum numbers of baryon charge , electric charge and strangeness , while @xmath0 , @xmath18 and @xmath19 denote the chemical potentials associated with the net densities of baryon number , @xmath20 , electric charge , @xmath21 , and strangeness , @xmath22 , respectively .
the particle number density @xmath23 of species @xmath11 follows from the thermodynamic identity @xmath24 as @xmath25 so that the above net densities are given by @xmath26 for @xmath27 . the actual conditions present in a heavy - ion collision are , however , more complex : first , the chemical potentials @xmath0 , @xmath18 and @xmath19 are not independent , but related to @xmath8 as well as to each other via the conditions @xmath28 here , the factor @xmath29 in eq .
( [ conditions ] ) accounts approximately for the ratio of protons to baryons in the colliding nuclei , while @xmath30 reflects the initial net - strangeness content .
second , during its expansion the created matter does not maintain chemical equilibrium , since the time scales for the necessary inelastic scatterings among the hadrons are typically much longer than the duration of the hadronic stage @xcite .
while hadrons are assumed to be formed in chemical equilibrium at the transition temperature @xmath1 , for not too small temperatures @xmath31 only the particle - number preserving interactions mediated by hadronic resonance decay and regeneration ( e.g. @xmath32 , @xmath33 , @xmath34 etc ) continue to occur with sufficient rate .
the hadronic matter is , therefore , in a state of partial chemical equilibrium @xcite , because the resonances are still in chemical equilibrium with their decay products , whereas the multiplicity ratios of hadrons , which are stable against strong decay during the hadronic stage , are frozen out at @xmath2
. the chemical potentials of the resonances @xmath35 become functions of the chemical potentials of all stable hadrons @xmath36 via @xmath37 and , consequently , the final particle number of a hadron species @xmath36 is given by @xmath38 , where the sum runs over all @xmath35 decaying into @xmath36 , @xmath39 and @xmath40 denote the primordial particle numbers of @xmath36 and @xmath35 , and @xmath41 gives the average number of @xmath36 produced in the decay of @xmath35 . as discussed in ref .
@xcite , this connection can easily be extended toward higher - order susceptibilities in order to study the average influence of resonance decays on the fluctuations in the final particle numbers . in this paper
we consider , in line with ref .
@xcite , a hrg model containing states up to a mass of @xmath42 gev as , for example , listed in the particle data book @xcite .
we consider as stable particles the mesons @xmath43 , @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 and @xmath5 and the baryons @xmath50 , @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 , @xmath56 and @xmath57 as well as their corresponding anti - baryons .
this implies that , as in the experimental analysis , feed - down from weak decays is explicitly excluded in our approach .
when applying the experimental acceptance cuts , we modify the hrg model integrals in the following way : @xmath58}{(-1)^{b_k+1}+\exp({(\sqrt{p_t^{\,2}{\rm cosh}[\eta]^2+m_k^2}-\mu_k)/t } ) } \end{gathered}\ ] ] in the case of cuts on the pseudo - rapidity @xmath5 ( for net - electric charge , i.e. for all charged particles ) and @xmath59}{(-1)^{b_k+1}+\exp({({\rm cosh}[y]\sqrt{p_t^{\,2}+m_k^2}-\mu_k)/t})}\end{gathered}\ ] ] in the case of cuts on the rapidity @xmath4 ( for net protons ) . due to the setup of the hrg model
, we can only apply cuts on the momenta of the primordial resonances and of the primordial stable hadrons , but not on those of the products of resonance decays .
it has , however , been estimated that the effect of acceptance cuts in rapidity on the decay daughters should be in the percent range @xcite .
furthermore the effect of radial flow on the measured particle species in a limited transverse momentum range , which might have been particularly important for the net charge since there a composite of particle species with different mass is considered , has been estimated taking into account the blastwave parameters extracted by star , and found to be negligible due to the rather wide @xmath6-window of the measurement ( 0.2 - 2 gev / c ) .
this result is in agreement with a similar previous study @xcite .
in the following , we compare our hrg model calculations with the efficiency corrected experimental results for the most central collisions ( @xmath60 ) published by the star collaboration for net - proton ) for net - protons can be found on the public star webpage . ]
@xcite and net - electric charge fluctuations @xcite .
the susceptibilities of conserved charges are defined as @xmath61 their relationship with the central moments of the conserved charge multiplicity distributions is @xmath62 where @xmath63 . from the quantities in eq .
( [ cenmom ] ) the following , to leading order volume - independent , ratios can be defined : @xmath64 we calculate the net - proton fluctuations according to the method presented in ref .
@xcite , where besides kinematic acceptance cuts also resonance decays and regeneration below the chemical freeze - out are taken into account . while resonance decays feed the distributions of the primordial protons and anti - protons , the regeneration of resonances leads to a randomization of the nucleon isospin : the dominant process is the regeneration of @xmath65-resonances from the scatterings of nucleons with thermal pions .
subsequently , these @xmath66-resonances decay into either the same or the opposite isospin state , where neutrons are , however , not detected experimentally .
consequently , additional fluctuations in the net - proton number arise , which we include based on the formalism by kitazawa and asakawa ( ka ) @xcite .
net - electric charge fluctuations are calculated based on the most abundant charged particles , namely pions , kaons , and protons as well as their anti - particles . also here , primordial distributions are fed by resonance decays , but corrections similar to the ka - corrections for the net - proton number are not needed , because processes via intermediate resonances conserve electric charge . while the application of the experimental acceptance cuts is straightforward in the net - proton case ( where @xmath29 gev / c @xmath67 gev / c and @xmath68 ) , it is more difficult in the case of net - electric charge . here , the general cuts are @xmath69 gev / c
@xmath70 gev / c and @xmath71 , but in order to suppress spallation protons , all protons ( and anti - protons ) with @xmath72 gev / c are removed in the experimental analysis . due to correlated resonance decay contributions to ( anti-)protons and pions or kaons , e.g. @xmath73 or @xmath74 , which are given by a single integral in the hrg model calculation , we can not cut the resonance contribution to the ( anti-)protons in the same way without also affecting the contributions to the pions and kaons .
we thus apply the lower @xmath6-cut of @xmath29 gev / c only to the primordial protons and anti - protons . in order to extract the freeze - out temperature and baryo - chemical potential for each collision energy , we have simultaneously analyzed two experimentally measured susceptibility ratios . with the resulting freeze - out conditions ( @xmath75 , @xmath76 ) we can calculate the remaining susceptibility ratios , which gives us a cross - check on the reliability of the determined freeze - out parameters .
the large experimental uncertainties in the higher - order susceptibility ratios of the net - electric charge @xmath77 and @xmath78 do not allow to meaningfully constrain the freeze - out temperature and baryo - chemical potential from net - electric charge fluctuations alone .
moreover , for the net protons , as already noted in ref .
@xcite , it is not possible to simultaneously reproduce @xmath79 and @xmath80 for all beam energies : this might point at a limitation in our approach , for example , due to an overestimate of the ka - corrections , which maximize the impact of isospin randomization .
several other effects that might impact the higher order moments have also not yet been considered , such as volume fluctuations @xcite , exact ( local ) charge conservation @xcite or repulsive van - der - walls forces among hadrons @xcite .
finally , the discrepancy in particular at larger @xmath0 , could also hint at the onset of chiral critical fluctuations in the higher moments ( skewness and above ) @xcite .
we therefore perform , first , a combined analysis of the ratios with the smallest experimental uncertainty , namely @xmath79 for net - electric charge and for net protons . in addition , we consider an alternative analysis using higher - order cumulants , namely @xmath79 for net - electric charge and @xmath80 for net protons , and discuss the difference in the extracted freeze - out parameters between these two choices .
( color online ) comparison between hrg model results and experimental data for the most central collisions ( @xmath60 ) ( from refs .
@xcite ) for @xmath79 of net - electric charge ( blue , upper symbols ) and net protons ( red , lower symbols ) .
the experimental data have been used in the hrg model in order to extract a freeze - out temperature and baryo - chemical potential for each collision energy.,scaledwidth=50.0% ] in fig .
[ fig1 ] , we show the experimental data as a function of collision energy per nucleon pair @xmath81 ( from refs .
@xcite ) together with our results for the first choice of fluctuation observables , i.e. the combined @xmath79 datasets .
we find that it is possible to extract , for each collision energy , a freeze - out temperature and baryo - chemical potential , which allow to simultaneously reproduce the ratios of the lowest - order susceptibilities for net protons and net - electric charge .
the smallest collision energy we consider is @xmath82 gev : below this energy we expect that the isospin randomization is not realized @xcite .
we note that for the determination of these freeze - out parameters the inclusion of the ka - corrections for @xmath79 of net protons , in accordance with ref .
@xcite , is essential .
freeze - out temperature ( upper panel ) and baryo - chemical potential ( lower panel ) as functions of the collision energy , extracted from the data in fig .
the corresponding values are listed in table [ tab1].,title="fig:",scaledwidth=90.0% ] freeze - out temperature ( upper panel ) and baryo - chemical potential ( lower panel ) as functions of the collision energy , extracted from the data in fig .
the corresponding values are listed in table [ tab1].,title="fig:",scaledwidth=90.0% ] in fig .
[ fig2 ] , we show the freeze - out temperature ( upper panel ) and baryo - chemical potential ( lower panel ) corresponding to this set of analyzed cumulant ratios , as functions of @xmath81 .
the precision in the experimental results allows a rather precise determination of these parameters .
the error bars shown in fig .
[ fig2 ] are based on hrg model calculations using the upper and lower uncertainty limits in the experimental data .
our values for @xmath2 are lower than those found in ref .
@xcite : even for the highest rhic energies , our results are close to the lower bound for @xmath1 determined in lattice qcd simulations @xcite .
this is evident in fig .
[ fig3 ] , where we show a comparison between the freeze - out points in the ( @xmath83 ) plane obtained in the present analysis and the curve of ref .
@xcite .
( color online ) freeze - out parameters in the ( @xmath83 ) plane : comparison between the curve obtained in ref .
@xcite ( red band ) and the values obtained from the combined analysis of @xmath79 for net - electric charge and net protons ( blue symbols ) presented here.,scaledwidth=50.0% ] using these freeze - out conditions , we now proceed to calculate the higher - order susceptibility ratios @xmath77 and @xmath78 for net protons and net - electric charge .
the results are shown in the different panels of fig .
[ fig4 ] in comparison with the experimental data .
it is evident that , with the obtained freeze - out conditions , one can reproduce all experimental results for the net - electric charge fluctuations ( left panels ) . as already mentioned , the agreement between our results and the experimental data for the net - proton @xmath80 becomes less accurate with decreasing collision energy ( upper right panel ) . for @xmath84 , our hrg model can not reproduce the anomalous depletion at the lower collision energies ( lower right panel ) , but is in good agreement with the data for the very low and very high @xmath81 ( notice that this depletion disappears in more peripheral collisions and can be described in central collisions by uncorrelated , i.e. independent , particle production when the experimentally determined proton and anti - proton distributions from star are used @xcite ) .
( color online ) comparison between hrg model results for @xmath85 and @xmath86 , with @xmath87 ( left ) and @xmath88 ( right ) as functions of @xmath81 ( blue crosses ) , and experimental data for the most central collisions ( @xmath60 ) from the star collaboration @xcite ( red diamonds ) .
the hrg model results are calculated for our new freeze - out conditions , listed in table [ tab1 ] . in all panels ,
acceptance cuts in the kinematics have been introduced , following the experimental analysis.,title="fig:",scaledwidth=50.0% ] ( color online ) comparison between hrg model results for @xmath85 and @xmath86 , with @xmath87 ( left ) and @xmath88 ( right ) as functions of @xmath81 ( blue crosses ) , and experimental data for the most central collisions ( @xmath60 ) from the star collaboration @xcite ( red diamonds ) .
the hrg model results are calculated for our new freeze - out conditions , listed in table [ tab1 ] . in all panels ,
acceptance cuts in the kinematics have been introduced , following the experimental analysis.,title="fig:",scaledwidth=50.0% ] ( color online ) comparison between hrg model results for @xmath85 and @xmath86 , with @xmath87 ( left ) and @xmath88 ( right ) as functions of @xmath81 ( blue crosses ) , and experimental data for the most central collisions ( @xmath60 ) from the star collaboration @xcite ( red diamonds ) .
the hrg model results are calculated for our new freeze - out conditions , listed in table [ tab1 ] . in all panels ,
acceptance cuts in the kinematics have been introduced , following the experimental analysis.,title="fig:",scaledwidth=50.0% ] ( color online ) comparison between hrg model results for @xmath85 and @xmath86 , with @xmath87 ( left ) and @xmath88 ( right ) as functions of @xmath81 ( blue crosses ) , and experimental data for the most central collisions ( @xmath60 ) from the star collaboration @xcite ( red diamonds ) .
the hrg model results are calculated for our new freeze - out conditions , listed in table [ tab1 ] . in all panels ,
acceptance cuts in the kinematics have been introduced , following the experimental analysis.,title="fig:",scaledwidth=50.0% ] ( color online ) freeze - out parameters in the ( @xmath83 ) plane : comparison between the values obtained by a combined analysis of @xmath79 for net - electric charge and net protons ( blue circles ) , and the values obtained from @xmath79 for net - electric charge and @xmath80 for net protons ( red squares).,scaledwidth=100.0% ] in order to determine by how much the freeze - out conditions need to be modified to reproduce the higher - order cumulants for the net protons , we perform , as second choice , a simultaneous analysis of @xmath79 for the net - electric charge and @xmath80 for the net protons .
the result improves the agreement with the measured @xmath80 , the values for @xmath79 of the net - proton distributions are , however , not described particularly well . for the net - electric charge fluctuation data the uncertainties are such that the alternate analysis is still within experimental error bars .
the comparison between the freeze - out parameters resulting from our two different calculations is shown in fig .
[ fig6 ] . while for high collision energies the two parameter sets are very similar , differences arise for smaller @xmath81 .
in order to be able to reproduce the higher - order cumulants , the curvature of the freeze - out curve turns up for larger baryo - chemical potential , which is opposite to lattice expectations .
this points at an inconsistency between the hrg model description of the lower- and higher - order cumulants in the net - proton distributions , which might signal the onset of chiral critical fluctuations in the higher order cumulants at large @xmath0 @xcite . in order to test whether
, in the net - proton analysis , already the second order cumulant could be affected , we have determined the @xmath89 ratio and found it to be consistent with unity for both , the hrg model and the data , for all measured collision energies .
in addition , since the gross features of the particle distributions are given by their lower - order cumulants , while higher - order cumulants are more sensitive to finer details , such as excluded volume effects or volume fluctuations , as well as to interactions in the late hadronic stage , obtaining chemical freeze - out parameters from the analysis of @xmath79 for net - electric charge and net - proton number is more reliable than using @xmath80 for the net - proton number .
the corresponding values for the freeze - out temperature and baryo - chemical potential for the different collision energies are given in table [ tab1 ] .
.[tab1 ] in this table we list the values of @xmath3 and @xmath2 at chemical freeze - out , corresponding to the relative collision energies . these values are based on our combined analysis of the data in fig . [ fig1 ] . [ cols="^,^,^",options="header " , ]
in conclusion , our study shows that we can simultaneously describe the net - electric charge fluctuations and the lower - order cumulants of the net - proton multiplicity distributions measured at rhic for collision energies spanning over more than an order of magnitude ( @xmath90 gev ) .
we calculated these fluctuation observables within the hrg model including the experimental acceptance cuts and the effects of resonance decays and regeneration . from a combined analysis of @xmath79 for net - electric charge and net - proton number ,
we obtain the freeze - out conditions summarized in table [ tab1 ] . given the reported uncertainties in the measured fluctuation observables , the resulting freeze - out temperatures are constrained to better than 5 mev for @xmath91 gev . with these freeze - out values ,
the higher - order susceptibility ratios for net - electric charge and net - proton number are reasonably well reproduced . if one takes the experimentally given particle samples as approximate representatives for the quantum numbers of electric and baryon charge , similar studies in lattice qcd yield a remarkable agreement for the collision energy dependence of @xmath2 and @xmath3 @xcite .
we note that a useful cross - check of our extracted chemical freeze - out parameters can be provided through the independent determination of the same parameters via a common fit of standard shms to experimental particle yields or ratios @xcite .
preliminary results of measured particle ratios from the rhic beam energy scan @xcite for all @xmath0 values analyzed here , yield , based on those standard shm fits , freeze - out temperatures ranging from ( @xmath92 ) mev for @xmath93 gev collisions when using a common fit for all particles ( including strange particles ) . at first glance , our parameters are below those extracted from the particle ratio fits as is also evident from fig .
[ fig3 ] . in order to quantify this difference
we show in fig .
[ fig7 ] a comparison of particle ratios obtained with our parameters and some standard shm parameters to the properly feed - down corrected particle ratios measured by star at the highest rhic collision energy @xcite .
( color online ) comparison between star particle ratio data for central events at @xmath81=200 gev @xcite and hrg model results for the specified chemical freeze - out parameters.,scaledwidth=90.0% ] the overall agreement with the data is roughly equivalent for both parameter sets .
our calculation significantly improves on the ( anti-)proton yield , but falls short for the strange baryons .
we note that the cumulant ratios of net charge and net protons are pion and proton dominated , which could be the reason that in our analysis the particle and fluctuation ratios yield consistently lower temperatures for the light quark sector .
recent lhc data @xcite seem to suggest a separation of chemical freeze - out parameters according to particle flavor , which is also supported by the latest lattice qcd simulations @xcite and sequential shms @xcite .
a fit to the strange baryon over pion ratios alone yields chemical freeze - out temperatures that are consistent with the standard shm fits @xcite . in order to determine the freeze - out parameters from fluctuation ratios in the strange sector we anticipate that at least efficiency corrected strange meson cumulants from the experiments will be available soon .
we gratefully acknowledge useful discussions with bill llope . this work is supported by the italian ministry of education , universities and research under the firb research grant rbfr0814tt , the hessian loewe initiative helmholtz international center for fair , and the us department of energy grants de - fg02 - 03er41260 , de - fg02 - 05er41367 and de - fg02 - 07er41521 . | we calculate ratios of higher - order susceptibilities quantifying fluctuations in the number of net protons and in the net - electric charge using the hadron resonance gas ( hrg ) model .
we take into account the effect of resonance decays , the kinematic acceptance cuts in rapidity , pseudo - rapidity and transverse momentum used in the experimental analysis , as well as a randomization of the isospin of nucleons in the hadronic phase . by comparing these results to the latest experimental data from the star collaboration
, we determine the freeze - out conditions from net - electric charge and net - proton distributions and discuss their consistency . |
with the advent of the internet , the exponential growth of the world - wide - web and routers confront people with an information overload @xcite .
we are facing too much data to be able to effectively filter out the pieces of information that are most appropriate for us .
a promising way is to provide personal recommendations to filter out the information .
recommendation systems use the opinions of users to help them more effectively identify content of interest from a potentially overwhelming set of choices @xcite . motivated by the practical significance to the e - commerce and society , various kinds of algorithms have been proposed , such as correlation - based methods @xcite , content - based methods @xcite , the spectral analysis @xcite , principle component analysis @xcite , network - based methods @xcite , and so on . for a review of current progress , see ref . @xcite and the references therein .
one of the most successful technologies for recommendation systems , called _ collaborative filtering _ ( cf ) , has been developed and extensively investigated over the past decade @xcite .
when predicting the potential interests of a given user , such approach first identifies a set of similar users from the past records and then makes a prediction based on the weighted combination of those similar users opinions . despite its wide applications , collaborative filtering suffers from several major limitations including system scalability and accuracy @xcite
recently , some physical dynamics , including mass diffusion @xcite , heat conduction @xcite and trust - based model @xcite , have found their applications in personal recommendations .
these physical approaches have been demonstrated to be of both high accuracy and low computational complexity @xcite
. however , the algorithmic accuracy and computational complexity may be very sensitive to the statistics of data sets .
for example , the algorithm presented in ref .
@xcite runs much faster than standard cf if the number of users is much larger than that of objects , while when the number of objects is huge , the advantage of this algorithm vanishes because its complexity is mainly determined by the number of objects ( see ref .
@xcite for details ) . in order to increase the system scalability and accuracy of standard cf
, we introduce a network - based recommendation algorithm with spreading activation , namely sa - cf .
in addition , two free parameters , @xmath0 and @xmath1 are presented to increase the accuracy and personality .
denoting the object set as @xmath2 and user set as @xmath3 = @xmath4 @xmath5 @xmath6 , a recommendation system can be fully described by an adjacent matrix @xmath7 , where @xmath8 if @xmath9 is collected by @xmath10 , and @xmath11 otherwise . for a given user ,
a recommendation algorithm generates a ranking of all the objects he / she has not collected before .
based on the user - object matrix @xmath12 , a user similarity network can be constructed , where each node represents a user , and two users are connected if and only if they have collected at least one common object . in the standard cf , the similarity between @xmath13 and @xmath10 can be evaluated directly by a correlation function : @xmath14 where @xmath15 is the degree of user @xmath13 . inspired by the diffusion process presented by zhou _
@xcite , we assume a certain amount of resource ( e.g. recommendation power ) is associated with each user , and the weight @xmath16 represents the proportion of the resource , which @xmath10 would like to distribute to @xmath13 . following a network - based resource - allocation process @xcite where each user distributes his / her initial resource equally to all the objects
he / she has collected , and then each object sends back what it has received to all the users collected it , the weight @xmath16 ( the fraction of initial resource @xmath10 eventually gives to @xmath13 ) can be expressed as : @xmath17 where @xmath18 denotes the degree object @xmath19 . using the spreading process ,
the user correlation network can be constructed , whose edge weight is obtained by eq .
( [ equation2 ] ) . for the user - object pair @xmath20 ,
if @xmath13 has not yet collected @xmath21 ( i.e. @xmath22 ) , the predicted score , @xmath23 , is given as @xmath24 from the definition of eq.([equation1 ] ) , one can get that , to a target user , all of his neighbors collection information would affect the recommendation results , which is different with the definition reachability @xcite . based on the definitions of @xmath16 and @xmath23 , sa - cf can be given .
the framework of the algorithm is organized as follows : ( i ) calculate the user similarity matrix @xmath25 based on the spreading approach ; ( ii ) for each user @xmath26 , obtain the score @xmath23 on every object not being yet collected by @xmath27 ; ( iii ) sort the uncollected objects in descending order of @xmath23 , and those in the top will be recommended .
to test the algorithmic accuracy and personality , we use a benchmark data - set , namely _ movielens _ @xcite .
the data consists of 1682 movies ( objects ) and 943 users , who vote movies using discrete ratings 1 - 5 .
hence we applied the coarse - graining method previously used in refs .
@xcite : a movie is set to be collected by a user only if the giving rating is larger than 2 .
the original data contains @xmath28 ratings , 85.25% of which are @xmath29 , thus the user - object ( user - movie ) bipartite network after the coarse gaining contains 85250 edges . to test the recommendation algorithms , the data set
is randomly divided into two parts : the training set contains 90% of the data , and the remaining 10% of data constitutes the probe . the training set is treated as known information , while no information in the probe set is allowed to be used for prediction .
a recommendation algorithm should provide each user with an ordered queue of all its uncollected objects .
it should be emphasized that , the length of queue should not be given artificially , because of the fact that the number of uncollected movies for different users are different .
for an arbitrary user @xmath13 , if the relation @xmath13-@xmath21 is in the probe set ( according to the training set , @xmath21 is an uncollected object for @xmath13 ) , we measure the position of @xmath21 in the ordered queue .
for example , if there are @xmath30 uncollected movies for @xmath13 , and @xmath21 is the 10th from the top , we say the position of @xmath21 is @xmath31 , denoted by @xmath32 . since the probe entries are actually collected by users , a good algorithm is expected to give high recommendations to them , thus leading to small @xmath33 .
therefore , the mean value of the position @xmath33 , @xmath34 ( called _ ranking score _
@xcite ) , averaged over all the entries in the probe , can be used to evaluate the algorithmic accuracy : the smaller the ranking score , the higher the algorithmic accuracy , and vice verse . implementing the sa - cf and cf @xcite ,
the average value of ranking score are @xmath35 and @xmath36 @xcite . clearly , under the simplest initial configuration , subject to the algorithmic accuracy ,
the sa - cf algorithm outperforms the standard cf . [ 1.0 ] vs. @xmath0 .
the black solid and red dash curves represent the performances of sa - cf and cf , respectively .
all the data points are obtained by averaging over ten independent runs with different data - set divisions.,title="fig : " ] [ 1.0 ] vs. @xmath0 .
the black solid , red dashed and green dotted curves represent the cases with typical length @xmath37 and 50 , respectively .
the blue dot line corresponds to the optimal value @xmath38 .
all the data points are obtained by averaging over ten independent runs with different data - set divisions.,title="fig : " ] [ 1.0 ] vs. @xmath0 .
the black solid , red dashed and green dotted curves represent the cases with typical lengths @xmath37 and 50 , respectively .
the blue dot line corresponds to the optimal value @xmath38 .
all the data points are obtained by averaging over ten independent runs with different data - set divisions.,title="fig : " ] [ 1.0 ] vs. @xmath1 .
the inset shows the relation for larger @xmath1 .
clearly , when @xmath1 approaches @xmath39 , the algorithmic accuracy is the same as that of the sa - cf with @xmath40 .
all the data points are obtained by averaging over ten independent runs with different data - set divisions .
, title="fig : " ]
in order to further improve the algorithmic accuracy , we propose two modified methods . similar to the ref .
@xcite , taking into account the potential role of object degree may give better performance .
accordingly , instead of eq .
( 2 ) , we introduce a more complicated way to get user - user correlation : @xmath41 where @xmath0 is a tunable parameter .
when @xmath42 , this method degenerates to the algorithm mentioned in the last section .
the case with @xmath43 weakens the contribution of large - degree objects to the user - user correlation , while @xmath44 will enhance the contribution of large - degree objects .
according to our daily experience , if two users @xmath13 and @xmath10 has simultaneously collected a very popular object ( with very large degree ) , it does nt mean that their interests are similar ; on the contrary , if two users both collected an unpopular object ( with very small degree ) , it is very likely that they share some common and particular tastes .
therefore , we expect a larger @xmath0 ( i.e. @xmath43 ) will lead to higher accuracy than the routine case @xmath42 .
fig.[fig1.1 ] reports the algorithmic accuracy as a function of @xmath0 .
the curve has a clear minimum around @xmath45 , which strongly support the above statement .
compared with the routine case ( @xmath42 ) , the ranking score can be further reduced by 11.2% at the optimal value .
it is indeed a great improvement for recommendation algorithms .
besides accuracy , the average degree of all recommended movies @xmath46 and the mean value of hamming distance @xmath47 @xcite are taken in account to measure the algorithmic personality .
the movies with higher degrees are more popular than the ones with smaller degrees
. the personal recommendation should give small @xmath46 to fit the special tastes of different users .
fig.[fig1.2 ] reports the average degree of all recommended movies as a function of @xmath0 .
one can see from fig.[fig1.2 ] that the average degree is negatively correlated with @xmath0 , thus depressing the recommendation power of high - degree objects gives more opportunity to the unpopular objects .
the hamming distance , @xmath48 , is defined by the mean value among any two recommended lists of @xmath13 and @xmath10 , where @xmath49 , @xmath50 is the list length and @xmath51 is the overlapped number of objects in the two users recommended lists .
fig.[fig1.3 ] shows the positively correlation between @xmath47 and @xmath0 , in according with the simulation results in fig.[fig1.2 ] , which indicates that depressing the influence of high - degree objects makes the recommendations more personal .
the above simulation results indicate that sa - cf outperforms cf from the viewpoints of accuracy and personality . besides the algorithmic accuracy and personality , the computational complexity should also be taken into account . actually , we argue that a better algorithm should simultaneously require less computation and generate higher accuracy .
note that , the computational complexity of eq .
( 3 ) is very high if the number of user , @xmath52 , is huge . actually , the majority of user - user similarities are very small , which contribute little to the final recommendation
. however , those inconsequential items , corresponding to the less similar users , dominate the computational time of eq . ( 3 ) . therefore , we propose a modified algorithm , so - called top-@xmath1 sa - cf , which only considers the @xmath1 most similar users information to any given user .
that is to say , in the top-@xmath1 sa - cf , the sum in eq .
( 3 ) runs over only the @xmath1 most similar users of @xmath13 . in the process of calculation
the similarity matrix @xmath16 , to each other , we can simultaneously record its most similar users . when @xmath53 , the additional computing time for top-@xmath1 similar users are remarkably shorter than what we can save from the traditional calculation of eq .
( 3 ) . more
surprisingly , as shown in fig.[fig1.4 ] , with properly chosen @xmath1 , this algorithm not only reduces the computation , but also enhances the algorithmic accuracy .
this property is of practical significance , especially for the huge - size recommender systems . from figures [ fig1.2 ] and [ fig1.3 ]
, one can find that , to the same @xmath0 range , the anticorrelations between @xmath46 , @xmath47 and @xmath0 are different in different @xmath0 range .
maybe there is a phase transition in the anticorrelations . because this paper mainly focuses on the accuracy and personality of the recommendation algorithms , this issue
would be investigated in the future .
in this paper , the spreading activation approach is presented to compute the user similarity of the collaborative filtering algorithm , named sa - cf .
the basic sa - cf has obviously higher accuracy than the standard cf .
ignoring the degree - degree correlation in user - object relations , the algorithmic complexity of sa - cf is @xmath54 , where @xmath55 and @xmath56 denote the average degree of users and objects . correspondingly , the algorithmic complexity of the standard cf is @xmath57 , where the first term accounts for the calculation of similarity between users , and the second term accounts for the calculation of the predictions . in reality , the number of users , @xmath52 , is much larger than the average object degree , @xmath56 , therefore , the computational complexity of sa - cf is much less than that of the standard cf .
the sa - cf has great potential significance in practice .
furthermore , we proposed two modified algorithms based on sa - cf .
the first algorithm weakens the contribution of large - degree objects to user - user correlations , and the second one eliminates the influence of less similar users .
both the two modified algorithm can further enhance the accuracy of sa - cf .
more significantly , with properly choice of the parameter @xmath1 , top-@xmath1 sa - cf can simultaneously reduces the computational complexity and improves the algorithmic accuracy .
a natural question on the presented algorithms is whether these algorithms are robust to other data - sets or random recommendation ?
to sa - cf , the answer is yes , because it would get the user similarity more accurately . while to the two modified algorithms ,
the answer is no . since both of the two modified algorithms introduced the tunable parameters @xmath0 and @xmath1 , the optimal values of different data - sets are different .
the further work would focus on how to find an effective way to obtain the optimal value exactly , then the modified algorithms could be implemented more easily .
we are grateful to tao zhou , zoltn kuscsik , yi - cheng zhang and met@xmath58 medo for their greatly useful discussions and suggestions .
this work is partially supported by sbf ( switzerland ) for financial support through project c05.0148 ( physics of risk ) , the national basic research program of china ( 973 program no .
2006cb705500 ) , the national natural science foundation of china ( grant nos . 60744003 , 10635040 , 10532060 , 10472116 )
, gq was supported by the liaoning education department ( grant no .
20060140 ) .
j. l. herlocker , j. a. konstan , k. terveen , and j. t. riedl , acm trans .
* 22 * , 5 ( 2004 ) .
j. a. konstan , b. n. miller , d. maltz , j. l. herlocker , l. r. gordon , and j. riedl , commun .
acm * 40 * , 77 ( 1997 ) .
m. balabanovi and y. shoham , commun .
acm * 40 * , 66 ( 1997 ) .
m. j. pazzani , artif .
intell . rev .
* 13 * , 393 ( 1999 ) .
d. billsus and m. pazzani , _ proc .
intl conf .
machine learning _
b. sarwar , g. karypis , j. konstan , and j. riedl , _ proc .
acm webkdd workshop _ ( 2000 ) .
k. goldberg , t. roeder , d. gupta , and c. perkins , inform .
ret . * 4 * , 133 ( 2001 ) .
zhang , m. blattner , and y .- k .
yu , phys .
99 * , 154301 ( 2007 ) .
c . zhang , m. medo , j. ren , t. zhou , t. li , and f. yang , epl * 80 * , 68003 ( 2007 ) .
t. zhou , j. ren , m. medo , and y .- c .
zhang , phys .
e * 76 * , 046115 ( 2007 ) .
t. zhou , l .- l . jiang , r .- q .
su , and y .- c .
zhang , epl * 81 * , 58004 ( 2008 ) .
g. adomavicius and a. tuzhilin , ieee trans .
know . & data eng . *
17 * , 734 ( 2005 ) .
z. huang , h. chen , and d. zeng , acm trans .
* 22 * , 116 ( 2004 ) .
b. sarwar , g. karypis , j. konstan , and j. reidl , in proceedings of the acm conference on electronic commerce .
acm , new york , 158 ( 2000 ) .
note that , the ranking score of the standard cf reported here is slightly different from that of the ref .
@xcite . it is because in this paper , if a movie in the probe set has not yet appeared in the known set , we automatically remove it from the probe ; while the ref .
@xcite takes into account those movies only appeared in the probe via assigning zero score to them .
another alterative way is to automatically move those movies from the probe set to the known set , which guarantees that any target movie in the probe set has been collected by at least one user in the known set .
the values of @xmath34 are slightly different for these three implementations , however , the choice of different implementations has no substantial effect on our conclusions . | in this paper , we propose a spreading activation approach for collaborative filtering ( sa - cf ) . by using the opinion spreading process
, the similarity between any users can be obtained .
the algorithm has remarkably higher accuracy than the standard collaborative filtering ( cf ) using pearson correlation .
furthermore , we introduce a free parameter @xmath0 to regulate the contributions of objects to user - user correlations .
the numerical results indicate that decreasing the influence of popular objects can further improve the algorithmic accuracy and personality .
we argue that a better algorithm should simultaneously require less computation and generate higher accuracy .
accordingly , we further propose an algorithm involving only the top-@xmath1 similar neighbors for each target user , which has both less computational complexity and higher algorithmic accuracy . |
in this paper we develop and analyze new methodology for inference about the mean of a stochastic process from data that consists of independent realizations of a stochastic process observed at discrete times , where each observation is contaminated by an additive error term .
formally , let @xmath0 be a stochastic process with mean function @xmath1\ ] ] and covariance function @xmath2 for all @xmath3 .
we denote the zero mean process @xmath4 by @xmath5 .
we observe @xmath6 at times @xmath7 , for @xmath8 , @xmath9 , that are of the form @xmath10 where @xmath11 , with mean @xmath12 , are random independent realizations of the process @xmath13 .
we assume that @xmath14 are independent across @xmath15 and @xmath16 with zero mean and variance @xmath17 = \sigma_{\eps}^2 $ ] .
+ in this paper we propose new methods for estimating and constructing confidence bands for @xmath18 . although the estimation of @xmath18 received considerable attention over the last decade , the theoretical study of data adaptive estimators in model ( [ model ] ) is still open to investigation .
in contrast with the abundance of methods for estimating @xmath18 , methods for constructing confidence bands for @xmath18 are very limited .
this motivates our twofold contribution to the existing literature : ( 1 ) we construct computationally efficient and fully data - driven estimators and confidence bands for @xmath18 , without making distributional assumptions on the process @xmath5 or smoothness assumptions on @xmath18 ; ( 2 ) we assess the quality of our data adaptive estimates theoretically and prove that both the estimators and the confidence bands adapt to the unknown regularity of @xmath18 .
moreover , we show that our bands are , asymptotically in @xmath19 , uniform in @xmath18 .
+ in what follows we review the existing results in the literature and provide further motivation for our procedure .
the problem of estimating @xmath18 from data generated from ( [ model ] ) has been considered by a large number of authors , starting with ramsay and silverman ( 2002 , 2005 ) and rupert , wand and carroll ( 2003 ) .
the existing methods are either based on kernel smoothers [ see , e.g. , zhang and chen ( 2007 ) , yao ( 2007 ) , benko , hrdle and kneip ( 2009 ) ] , penalized splines [ see , e.g. , ramsay and silverman ( 2005 ) ] , free - knot splines [ see , e.g. , gervini ( 2006 ) ] , or ridge - type least squares estimates , [ see , e.g. , rice and silverman ( 1991 ) ] .
all resulting estimates depend on tuning parameters that are method specific
. theoretical properties of these estimates of @xmath18 are still emerging , and have only been established for non - adaptive choices of the respective tuning parameters , that is choices that require prior knowledge of the smoothness of @xmath18 , [ see , e.g. zhang and chen ( 2007 ) and gervini ( 2006 ) ] .
although guidelines for data - driven choices of these parameters are offered in all these works , the theoretical properties of the resulting estimates are still open to investigation . in contrast , we suggest in section 2 below a computationally simple method based on thresholded least squares estimators .
our method does not require any specification of the regularity of @xmath12 or @xmath13 prior to estimation .
we show via oracle inequalities that our estimators adapt to this unknown regularity . + whereas the estimation of the mean @xmath12 of the process @xmath13 is well understood , modulo the technical and possibly computational issues raised above , the construction of uniform confidence intervals for @xmath18 has not been investigated in this context and in general the construction of confidence bands for @xmath18 in model ( [ model ] ) seems to have received little attention .
zhang and chen ( 2007 ) and yao ( 2007 ) construct kernel - type estimators and show that they are asymptotically normal with mean @xmath12 and variance @xmath20 .
although not addressed implicitly in these works , one can use these results to build confidence bands .
this construction would require the estimation of @xmath21 , using for instance the large body of work on the estimation of the covariance operator , based on the karhunen - loeve decomposition of the process @xmath13 and the subsequent estimation the the functional eigenfunctions and eigenvalues , see , e.g. , mller ( 2005 ) , benko , hrdle and kneip ( 2009 ) , who also comment on the possible instability of these estimates and offer refined methods for improved performance .
we offer an alternative method in section 2.5 below .
our procedure is computationally simple , avoids direct estimation of the covariance matrix @xmath22 , @xmath23 , and leads to adaptive bands that are uniform in the parameter @xmath18 .
+ the rest of the paper is organized as follows . in section 2.2 below we discuss thresholded least squares estimators in the functional data setting .
our emphasis is on hard threshold estimators , but we also discuss briefly the closely related soft threshold estimators . in section 2.3
we establish oracle inequalities for the fit of the estimators which show that the estimates adapt to the unknown sparsity of the mean @xmath18 .
the sparsity of @xmath18 is relative to a given approximating basis . in section 2.4 we suggest cross - validation for choosing the basis from a library of bases . since each basis induces an estimator , our procedure can be regarded as one of selecting an estimator from a given list .
we then establish an oracle inequality for the selected estimator , that shows that the selected estimator performs , essentially , as well as the best estimator from the list , in terms of mean squared error relative to the unknown @xmath18 . in section 2.5
we give the construction of the confidence bands and prove that they have the desired coverage probability .
section 3 contains a comprehensive simulation study that strongly supports the theoretical merits of the method and indicates that our method compares favorably with existing methods .
the net merit of the proposed method is very visible when the variance of the random noise @xmath24 is at the same level as that of the stochastic process @xmath5 and we discuss this in detail in sections 3.2 and 3.3 .
all the proofs are collected in the appendix .
as explained in the introduction , the aim of this paper is ( a ) to estimate the mean @xmath12 of the process @xmath13 and ( b ) to construct confidence bands for the mean @xmath12 .
our approach is based on thresholded least squares estimates obtained relative to bases @xmath25 that are orthonormal in @xmath26 , where @xmath27 is the empirical measure that puts mass @xmath28 at each @xmath29 .
thus , our bases satisfy @xmath30 for @xmath31 .
examples include the fourier , local trigonometric and haar bases .
+ since @xmath25 is orthonormal in @xmath26 , each @xmath32 has the decomposition @xmath33 with @xmath34 and @xmath35 for ease of notation we suppress the dependence on @xmath36 in @xmath37 and @xmath38 . under model ( [ model ] ) the random variables @xmath39 , for each @xmath40 , are independent and identically distributed with mean zero and variance @xmath41 = \frac{1}{m^2 } \sum_{j=1}^m \sum_{j'=1}^m \gamma ( t_j , t_{j ' } ) \phi_k(t_j ) \phi_k(t_{j ' } ) ; \end{aligned}\ ] ] in the special case where @xmath42 for all @xmath43 , the variances @xmath44 reduce to @xmath45 for @xmath46 .
the coefficients @xmath37 determine the target vector @xmath47 via the formula @xmath48 for each @xmath16 .
we motivate below our proposed methods for inference on @xmath47 .
+ our procedure falls between two of the currently used strategies : averaging estimated individual trajectories and applying various smoothing methods to the entire data set .
our initial estimator of @xmath49 is a least squares estimator , which can be viewed as an average ( over @xmath19 ) of weighted values of the @xmath6 s .
our final estimator will be a truncated version of the least squares estimator , with data dependent truncation levels determined from the entire data set .
we describe our procedure below .
the least squares estimator based on all observations of @xmath50 , for @xmath37 defined by ( [ mu ] ) , is the vector @xmath51 that minimizes @xmath52 over @xmath53 . using the orthonormality property of the basis
the estimators @xmath54 of @xmath37 are given by @xmath55 the sample average ( over @xmath19 ) of @xmath56 , which in turn are the least squares estimators of @xmath57 based on the observations @xmath6 from the @xmath15-th curve only .
using again the orthonormality property of the basis used for this fit , the estimators @xmath58 of @xmath59 are given by @xmath60 recalling that each @xmath6 follows model ( [ model ] ) , and using the representations ( [ xi ] ) ( [ aik ] ) , we can further write @xmath56 as @xmath61 since @xmath14 and @xmath38 have mean zero and are independent across @xmath15 and @xmath16 we obtain , for every @xmath40 , that @xmath62=\mu_k \ \ \mbox{and } \ \ \text{var}(\wh\mu_{i , k})=\sigma_k^2 + \frac{\sigma_\eps^2 } { m}.\ ] ] similarly , we find that @xmath63 with @xmath64=\mu_k \ \
\mbox{and } \ \
\text{var}(\wh\mu_{k})=\frac{\sigma_k^2}{n } + \frac{\sigma_\eps^2 } { mn}.\ ] ] the initial ( unbiased ) estimator @xmath65 based on the least squares estimates @xmath54 of the mean function @xmath49 is simply @xmath66 and its variance may be unnecessarily inflated by the presence of , possibly many , very small estimates @xmath67 .
this can be remedied by truncating the coefficients at a level that takes into account both the variability of the measurement errors @xmath68 and the variability of the stochastic process @xmath5 ; this is the essential difference between truncated estimators based on data generated as in ( [ model ] ) and their counterpart based only on independent data in a standard nonparametric regression setting .
we will use the truncation level @xmath69 given below and we will justify it theoretically and practically in the next sections .
let @xmath70 be the quantile corresponding to a @xmath71 random variable and @xmath72 is a given number .
define @xmath73 for some small @xmath74 ( that is set to zero in practice ) that depends on @xmath75 notice that @xmath76 a consistent and unbiased estimator of @xmath77 since the @xmath58 , @xmath8 are i.i.d . for fixed @xmath78 .
+ we will focus on hard threshold estimators of the coefficients @xmath37 and function @xmath18 .
they are , respectively @xmath79 we will also consider , for completeness , the soft threshold estimators : @xmath80 the two estimators are closely related as the coefficients @xmath81 and @xmath82 differ by at most @xmath69 since @xmath83 [ [ oracle - inequalities - for - the - estimators - of - the - mean - of - a - stochastic - process ] ] oracle inequalities for the estimators of the mean of a stochastic process ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ to discuss the quality of the estimates given above relative to the mean @xmath18 , we first investigate their properties relative to a truncated version of @xmath18 and obtain the desired results as a consequence .
we motivate the truncation of @xmath18 below .
consider the theoretical truncation level @xmath84 where @xmath70 is the quantile corresponding to a @xmath71 random variable and @xmath72 is a given number .
notice that this is the population counterpart of the data based levels given in ( [ trunc ] ) above .
define , for each @xmath78 , @xmath85 and we write @xmath86 for a truncated version of @xmath18 . when the truncation @xmath87 retains the main features of @xmath18 , it can be considered as the new target for inference .
this is the approach we take in the sequel . as an illustrative example , we consider the mean function [ gaussian_mix ] @xmath88 shown in black in figures [ sparsity](a ) and [ sparsity](c ) .
we project @xmath89 onto linear subspaces in @xmath90 generated by the fourier and haar basis functions , respectively , evaluated at @xmath91 equally distant points in @xmath92 $ ] .
many of the projection coefficients @xmath93 are close to zero for both bases .
we consider the truncation level @xmath94 given above with @xmath95 , @xmath96 , @xmath97 and for @xmath98 given by ( [ var ] ) above corresponding to the brownian bridge process with covariance function @xmath99 .
+ figures [ sparsity](b ) and [ sparsity](d ) show @xmath100 versus their index , for the fourier and haar basis respectively . the reconstruction @xmath101 is shown in red in figures [ sparsity](a ) and [ sparsity](c ) , and is very close to @xmath12 in both cases .
notice that only 11 coefficients are needed for this good reconstruction of @xmath18 via the fourier basis , versus 92 via the haar basis , as we reconstruct a differentiable function with differentiable and non - differentiable basis functions , respectively .
following standard terminology in non - parametric estimation , we refer to the fact that @xmath18 can be reconstructed well via a smaller subset of the given collection of the basis functions by saying that @xmath18 has a sparse representation relative to that basis .
+ in what follows we show that the thresholded estimates introduced in the section above adapt to the sparsity of @xmath18 . of course
, since @xmath18 is unknown , so is its sparsity relative to a given basis .
nevertheless , we show that our estimators adapt to this unknown sparsity in terms of their fit , and refer to these results as oracle inequalities .
the type of oracle inequalities that we establish below illustrate that the fit of our estimators depends only on the estimation errors induced by the estimates of the non - zero coefficients of a sparse representation of @xmath18 within a given basis . as our example indicates , and as is the case in any non - parametric estimation problem , the overall quality of our estimator will further depend on the choice of the basis used for estimation .
we will therefore complement the construction of our estimator with a basis selection step .
we begin by stating our results for a given basis .
+ for both hard and soft threshold estimators we obtain estimation bounds on the fit at the observation points @xmath29 .
we formulate our results in terms of the empirical supremum norm @xmath102 and @xmath103 norm @xmath104 defined below . for any real function @xmath105 ,
let @xmath106 all theorems and results of this article require that @xmath107<\infty$ ] , for all @xmath108 , and that @xmath109<\infty$ ] .
the next three theorems are proved for a given basis and the desired probability @xmath110 .
all estimates are based on the threshold level @xmath111 given in ( [ trunc ] ) , for a user specified value of @xmath110 .
the following result establishes oracle inequalities for the hard - threshold estimators .
define @xmath112 which differs by @xmath113 from @xmath94 defined above in ( [ theolevel ] ) , for a quantity @xmath114 that is arbitrarily close to zero ; this is needed for purely technical reasons , and for all practical purposes @xmath115 and @xmath94 can be considered the same .
theorems [ thm1 ] and [ thm2 ] yield immediately results on the performance of these estimators relative to the untruncated @xmath18 .
in particular , the following inequality holds with probability at least @xmath123 as @xmath124 : @xmath125 for both estimators @xmath126 and @xmath127 .
this follows directly from the above results and the triangle inequality .
the first term @xmath128 can be viewed as the approximation error or bias term , whereas the second term represents the estimation error or standard deviation term .
the bias term is unavoidable , and its size depends on the basis choice .
it suggests the need for an adaptive method , that would select the basis that is best suited for the unknown underlying mean function @xmath18 .
we discuss this in section 2.4 below .
+ theorems [ thm1 ] and [ thm2 ] are novel type of oracle inequalities for thresholded estimators , as they guarantee the in probability " , rather than on average " , performance of the estimator , at any probability level of interest @xmath72 .
these properties hold for our estimates , as they are constructed relative to variable threshold levels that depend on @xmath110 . to the best of our knowledge ,
such results are new in the functional data context .
they are also new in the general non - parametric settings , where a more traditional way to state the oracle properties of the estimators is in terms of the expected mean squared error , see , for instance , donoho and johnstone ( 1995 , 1998 ) , wasserman ( 2006 ) , tsybakov ( 2009 ) and the references therein . for completeness
, we also give an assessment of our estimates in terms of the expected mean squared error in theorem [ thm3 ] below , which restates theorem [ thm1 ] in terms of expected values .
to avoid technical clutter , we consider the toy estimator @xmath129 in lieu of @xmath130 .
recall the notation @xmath131 theorem [ thm3 ] shows that the expected mean squared error of our estimator also adapts to the unknown sparsity of @xmath18 , as indicated by the first term in either inequality ( [ gen ] ) or ( [ normal ] ) .
the second term in these inequalities is essentially an average of the quantities that constitute the first term .
this is more evident from the closed form expression ( [ normal ] ) , and shows that this second term is negligible relative to the first one , especially for small values of @xmath110 .
the results of the previous section make it clear that the basis choice influences both the bias and the variance of our estimates ; the type of basis one uses for the fit can be regarded as the tuning parameter of our estimation procedure .
we give below a data adaptive procedure of selection and show in theorem [ ds ] below that the estimator based on the selected basis behaves essentially as if the best basis for approximating the unknown @xmath18 was known in advance .
+ we select the basis via a cross - validation ( data - splitting ) technique , by randomly dividing the @xmath19 discretized curves @xmath136 in two equally sized groups . the first sample @xmath137 is used for constructing various estimates , say @xmath138 , @xmath139 , based on various bases , choices of @xmath110 and thresholding methods ( hard and soft ) .
the second sample ( hold - out or validation sample ) @xmath140 is used to select the optimal estimate @xmath141 that minimizes the empirical risk @xmath142 over @xmath143 . here
@xmath144 is the index set for the curves that are set aside to evaluate the estimators @xmath138 and @xmath145 is its cardinality .
theorem [ ds ] requires that the process @xmath5 and the random error @xmath24 have moments strictly larger than 2 , which is still a very mild assumption .
+ the last term in the right hand side of ( [ sel ] ) is of order @xmath152 , making the sum of the last two terms of order @xmath152 .
this can be regarded as the price to pay for using a data adaptive procedure to select the appropriate basis .
the factor 2 multiplying the right hand side of ( [ sel ] ) can be reduced to @xmath153 at the cost of increasing the last two terms on the right by a factor proportional to @xmath154 , for @xmath155 arbitrarily close to zero . to avoid notational clutter we opted for using the constant 2 .
therefore , theorem [ ds ] shows that the basis selection process yields an estimate that is essentially as good as the best estimate on the list , in terms of expected squared error . since which is best can not be known in advance , as @xmath18 is unknown , the result of theorem [ ds ] can also be regarded as an oracle inequality . in this section
we will construct confidence bands for @xmath18 that are uniform over the parameter space .
we begin with the confidence band based on a hard threshold estimator given below . set @xmath156 and notice that it differs by @xmath114 from @xmath69 given in ( [ trunc ] ) above .
this is again needed for technical reasons , as in practice @xmath114 can be set to zero .
[ thm5 ] ( 1 ) for all @xmath116 , @xmath117 and @xmath74 @xmath157 contains @xmath158 with probability at least @xmath123 , as @xmath124 .
+ ( 2 ) moreover , if all non - zero coefficients @xmath37 exceed @xmath159 , the band can be made smaller by a factor 3 : @xmath160 contains @xmath161 with probability at least @xmath123 , as @xmath124 .
michal benko , wolfgang hrdle and alois kneip .
common functional principal components . _
annals of statistics _ 37(1 ) , 1 34 , 2009 .
+ david donoho and iain johnstone .
adapting to the unknown smoothness via wavelet shrinkage .
_ journal of the american statistical association _ 90 , 12001224,1995 .
+ david donoho and iain johnstone .
minimax estimation via wavelet shrinkage .
_ annals of statistics _ 26 , 789921 , 1998 .
+ christopher genovese and larry wasserman .
adaptive confidence bands . _
annals of statististics _
36(2 ) , 875905 , 2008 . + daniel gervini .
free - knot spline smoothing for functional data .
_ journal of the royal statistical society , series b _
68(4 ) , 671687 , 2006 . +
hans - georg mller . functional modelling and classification of longitudinal data .
_ scandinavian journal of statistics _ 32 , 223240 , 2005 .
+ hans - georg mller , rituparna sen and ulrich stadtmller .
functional data analysis for volatility .
_ manuscript _ , 2006 .
+ ` r ` development core team ( 2008 ) . `
r ` : _ a language and environment for statistical computing . _ ` r ` foundation for statistical computing , vienna , austria .
isbn 3 - 900051 - 07 - 0 , url http://www.r-project.org .
+ james ramsay and bernard silverman .
_ functional data analysis _ , 2@xmath302 edition .
springer , new york , 2005 .
+ james ramsay and bernard silverman .
_ applied functional data analysis . _
springer , new york , 2002 .
+ john rice and bernard silverman . estimating the mean and covariance structure nonparametrically when the data are curves .
_ journal of the royal statistical society , series b _
53(1 ) , 233243 , 1991 .
+ david ruppert , simon sheather and matthew wand .
an effective bandwidth selector for local least squares regression
. _ journal of the american statistical association _
90(432 ) , 1257 - 1270 , 1995 . + david ruppert , matthew wand and raymond carroll .
_ semiparametric regression . _ cambridge university press , cambridge 2003 .
+ burkhart seifert , michael brockmann , joachim engel and theo gasser .
fast algorithms for nonparametric curve estimation .
_ journal of computational and graphical statistics _
3(2 ) , 192213 , 1994 . + alexandre b. tsybakov . _
introduction to nonparametric estimation . _ springer , new york , 2009 .
+ larry wasserman .
_ all of nonparametric statistics .
_ springer , new york , 2006 .
+ marten wegkamp .
model selection in nonparametric regression .
_ annals of statistics
_ 31(1 ) , 252273 , 2003 .
+ fang yao .
asymptotic distributions of nonparametric regression estimators for longitudinal and functional data .
_ journal of multivariate analysis _ 98 , 4056 , 2007 .
+ fang yao , hans - georg mller and jane - ling wang .
functional data analysis for sparse longitudinal data .
_ journal of the american statistical association _
100(740 ) , 577590 , 2005 .
+ jin - ting zhang and jianwei chen . statistical inferences for functional data .
_ annals of statistics
_ 35(3 ) , 10521079 , 2007 . | this paper proposes and analyzes fully data driven methods for inference about the mean function of a stochastic process from a sample of independent trajectories of the process , observed at discrete time points and corrupted by additive random error .
the proposed method uses thresholded least squares estimators relative to an approximating function basis .
the variable threshold levels are estimated from the data and the basis is chosen via cross - validation from a library of bases .
the resulting estimates adapt to the unknown sparsity of the mean function relative to the selected approximating basis , both in terms of the mean squared error and supremum norm .
these results are based on novel oracle inequalities .
in addition , uniform confidence bands for the mean function of the process are constructed .
the bands also adapt to the unknown regularity of the mean function , are easy to compute , and do not require explicit estimation of the covariance operator of the process .
the simulation study that complements the theoretical results shows that the new method performs very well in practice , and is robust against large variations introduced by the random error terms .
+ keywords : stochastic processes ; nonparametric mean estimation ; thresholded estimators ; functional data ; oracle inequalities ; adaptive inference ; uniform confidence bands .
= 1 |
a _ calabi yau metric _ is a riemannian metric on a manifold @xmath1 of dimension @xmath2 whose riemannian holonomy is precisely @xmath3 .. ] the representation theory of the holonomy group gives rise to various geometric structures on @xmath1 : there is a compatible complex structure ( unique up to complex conjugation when @xmath4 ) , a @xmath5-form @xmath6 which serves as the khler form of the given metric with respect to that complex structure , and a nowhere - vanishing holomorphic @xmath7-form @xmath8 .
both @xmath6 and @xmath8 are covariantly constant ( as is the almost - complex structure operator @xmath9 ) .
thanks to a result conjectured by calabi @xcite and proven by yau @xcite , when @xmath1 is a compact khler manifold with a nowhere - vanishing holomorphic @xmath7-form , there is a unique ricci - flat metric in each de rham cohomology class @xmath10\in h^2(x,\mathbb r)$ ] containing a khler form .
these metrics have holonomy contained in @xmath3 , so they will be calabi
yau under our definition provided that the holonomy does not reduce to a subgroup . as a consequence , the existence of a nowhere - vanishing holomorphic @xmath7-form on a khler manifold @xmath1 is often taken as a definition of _ calabi
yau manifold_. given a calabi
yau metric on @xmath1 , an associated khler form @xmath6 , and a holomorphic @xmath7-form @xmath8 , we say that a submanifold @xmath11 of ( real ) dimension @xmath7 is _ special lagrangian _
if @xmath12 and @xmath13 for some @xmath14 called the _ phase _ of @xmath11 .
this notion was introduced by harvey and lawson @xcite as a key example of a _ calibrated geometry _ : such submanifolds have a local volume - minimizing property .
unfortunately , very few examples of special lagrangian submanifolds are known in the compact case .
let @xmath1 be a compact calabi
yau manifold .
physical arguments predict that for most such @xmath1 , there should exist a mirror partner @xmath15 , which is another compact calabi
yau manifold whose physical ( but not geometrical ) properties are closely related to those of @xmath1 .
however , as has been recognized since the early days of mirror symmetry @xcite , @xmath1 is expected to have a mirror partner only when the complex structure of @xmath1 is sufficiently close to a `` large complex structure limit point , '' which is a class of boundary points in the compactified moduli space @xmath16 characterized by having maximally unipotent monodromy @xcite .
strominger , yau , and zaslow @xcite argued on physical grounds that any such calabi yau manifold should have a map @xmath17 whose general fiber @xmath18 is a special langrangian @xmath7-torus @xmath19 ; such a structure is called a _
supersymmetric torus fibration of @xmath1 _ , since the special lagrangian condition is the geometric counterpart to the preservation of half of the supersymmetry in a physical model .
strominger , yau , and zaslow also proposed that the mirror partner should be given , to first approximation , by a compactification of the family of dual tori @xmath20 .
it is worth saying a few words about the physical construction of the mirror partner .
strominger , yau , and zaslow argue that the original calabi
yau manifold and its mirror partner should be related by a physical construction known as `` t - duality on the @xmath7-torus fibers . '' in the absence of holomorphic disks with boundaries on the @xmath7-tori , this t - duality simply replaces each nonsingular torus by its dual torus ( cf .
@xcite ) , while doing something unknown at the singular fibers .
this description is expected to be modified when holomorphic disks are present , but the precise effect of the holomorphic disks has not yet been worked out . and as we shall see , the current expectation ( at least when @xmath21 ) is that such holomorphic disks will be present for at least some of the tori in the torus fibration .
this has made it difficult to formulate a mathematical version of the original strominger
zaslow proposal which is both precise and accurate .
because the arguments used by strominger , yau , and zaslow implicitly assume that the calabi
yau metric is uniformly large , we put that hypothesis in the following version of their existence conjecture . for any calabi
yau metric on a compact complex manifold @xmath1 of complex dimension @xmath7 whose complex structure is sufficiently close to a large complex structure limit point and whose khler class is sufficiently deep in the khler cone , there exists a supersymmetric torus fibration @xmath17 , where @xmath22 is a homology @xmath7-sphere .
there is by now considerable indirect evidence in favor of this conjecture , including an explicit construction in a ( slightly degenerate ) limiting case @xcite , as well as two strategies ( @xcite and @xcite ) for constructing weak forms of these fibrations for a certain class of compact calabi
yau threefolds .
however , it has become clear that proving this conjecture will require developing new techniques for studying special lagrangian submanifolds of a calabi
yau manifold . in spite of our lack of tools to prove the conjecture , though , many qualitative features of supersymmetric torus fibrations have been inferred in various ways , and this paper is devoted to explaining our best current ( conjectural ) understanding of those qualitative features .
we introduce the following terminology and notation . given a supersymmetric torus fibration @xmath17 , we let @xmath23 be the set of _ singular points of fibers _ of @xmath24 , and let @xmath25 be the _ discriminant locus _ of the fibration .
in low dimension , supersymmetric torus fibrations of compact calabi yau manifolds are completely understood . in the case of elliptic curves ( @xmath26 ) , any calabi yau metric is flat and a special lagrangian @xmath27-torus is just a closed geodesic . as is well known ,
if the homology class is fixed then there is a fibration of the elliptic curve over @xmath28 by closed geodesics in the specified class , with no singular fibers . in the case of k3 surfaces ( @xmath29 ) ,
an analysis is possible due to the non - uniqueness of the compatible complex structure .
in fact , the original paper of harvey and lawson @xcite showed that if @xmath30 is special lagrangian , then there is a different complex structure on @xmath1 ( compatible with the given calabi
yau metric ) such that @xmath30 is a complex submanifold .
thus , a special lagrangian @xmath31 fibration can be interpreted in another complex structure as a holomorphic elliptic fibration @xmath32 ( with base @xmath33 ) , and the structure of these is known in detail .
( in fact , a generic ricci - flat metric on a k3 surface admits such a fibration @xcite , so the existence conjecture holds in this case . )
thanks to work of kodaira @xcite , a complete classification of possible singular fibers of such fibrations is known : they are characterized by the conjugacy class of the monodromy action on @xmath34 .
the simplest fibers , called _
semistable _ , are associated to unipotent monodromy transformations . in an appropriate basis ,
the monodromy matrix takes the form @xmath35 the topology of a semistable degeneration with @xmath36 is very familiar .
one of the cycles on the two - torus extends over the degeneration , and the other `` vanishing '' cycle shrinks to a point ; if we follow the torus around a loop encircling the degeneration point in the base , there is a dehn twist along the vanishing cycle . in spite of this twisting of the topology of the torus ,
though , the total space of the fibration is non - singular .
( in the physics literature , the corresponding geometry is known as the `` taub - nut metric . '' ) for the generic elliptic fibration of a k3 surface , all fibers are semistable , and there are exactly @xmath37 of them , each with @xmath36 .
the monodromy data for such a generic fibration ( choosing an arbitrary base point @xmath38 ) gives a natural homomorphism @xmath39 whose generating loops all map to matrices conjugate to eq . .
the mirror partner of a given k3 surface ( with a fixed ricci - flat metric ) is known to be another k3 surface with a different ricci - flat metric @xcite . when passing to the mirror ,
the monodromy matrices @xmath40 are replaced by @xmath41 ; since @xmath41 is conjugate to @xmath40 , the monodromy data does not change .
we will conjecturally extend this kind of `` topological '' mirror symmetry statement to dimension @xmath42 in the next section .
note that , in any dimension , if we replace all nonsingular tori by their dual tori , the monodromy matrices change as @xmath43 .
the first step in studying supersymmetric @xmath0 fibrations of compact calabi
yau threefolds is to study more general @xmath0 fibrations , without imposing the `` special lagrangian '' condition . for calabi
yau hypersurfaces in toric varieties ( of arbitrary dimension ) , zharkov @xcite constructed a ( topological ) @xmath19 fibration .
a general program to understand such fibrations @xmath32 ( in dimension @xmath42 ) for which the map @xmath24 is smooth ( i.e. , @xmath44 ) was initiated by gross @xcite , and parallel results were obtained by ruan @xcite in his study of a specific class of @xmath0 fibrations of calabi yau hypersurfaces .
the monodromy of such fibrations and the topology of the singular fibers was determined under a suitable assumption of genericity , analogous to the assumption of `` generic elliptic fibration '' in the case of k3 surfaces which guaranteed that all fibers were semistable with @xmath36 .
we can summarize the analysis in a general conjecture ( which conjecturally extends their results to the general case ) .
[ con : topology ] let @xmath17 be a smooth @xmath0 fibration which is smooth , not the fibers of the fibration . ] of a compact calabi
yau threefold which is generic in a suitable sense .
then * the discriminant locus of the fibration is a trivalent graph @xmath45 . *
the topology near the edges of @xmath45 is modeled by the product of a cylinder with the @xmath36 semistable degeneration of two - tori . in particular ,
a dehn twist along the vanishing cycle and a nonsingular total space are features of this topology .
* for any loop around an edge of @xmath45 , the monodromy on either @xmath46 or @xmath47 of the @xmath42-tori is conjugate to @xmath48 in particular , both monodromy actions have a @xmath5-dimensional fixed plane .
* the vertices of @xmath45 come in two types : near a _ positive vertex _ , the three monodromy actions on @xmath46 near the vertex have fixed planes whose intersection is @xmath27-dimensional , while the three monodromy actions on @xmath47 have a common @xmath5-dimensional fixed plane . in an appropriate basis ,
the monodromy matrices on @xmath47 take the form @xmath49 on the other hand , near a _ negative vertex _ , the three monodromy actions on @xmath46 near the vertex have a common @xmath5-dimensional fixed plane , while the three monodromy actions on @xmath47 have fixed planes whose intersection is @xmath27-dimensional .
in an appropriate basis , the monodromy matrices on @xmath47 take the form @xmath50 * the fiber of @xmath24 over any point of @xmath22 other than a vertex of @xmath45 has a fixed point free @xmath51 action and in particular has euler characteristic @xmath52 .
the fiber of @xmath24 over each positive vertex has euler characteristic @xmath27 , and the fiber over each negative vertex has euler characteristic @xmath53 .
( in fact , gross @xcite and ruan @xcite gave explicit descriptions of these singular fibers , but we will not reproduce those descriptions here . ) there is an induced global monodromy action on @xmath47 @xmath54 whose generators satisfy the conditions spelled out in the conjecture .
if we have a compact calabi yau threefold and its mirror partner , with smooth @xmath0 fibrations whose nonsingular fibers are dual to each other , then the monodromy transformations will be related as @xmath55 .
this implies that the rles of the positive and negative vertices in the fibration are reversed between a calabi
yau threefold and its mirror partner .
since part ( e ) of the conjecture implies that the topological euler number of @xmath1 can be calculated via @xmath56 the effect of mirror symmetry on monodromy then shows that the euler number changes sign : @xmath57 as expected from physical mirror symmetry arguments . in @xcite
, gross showed how to go further , and use the data from a generic smooth @xmath0 fibration of a given compact calabi
yau threefold to construct a manifold which is a candidate mirror partner .
the transpose inverse of the original monodromy representation produces a mirror monodromy representation , describing the monodromy on the family of dual tori ( with positive and negative vertices reversed ) .
gross proved a `` reconstruction theorem : '' the family of dual tori can be completed to a compact topological manifold with a smooth @xmath0 fibration , satisfying the properties stated in the conjecture .
for smooth @xmath0 fibrations of a compact calabi
yau threefold , the combinatorics of the graph @xmath45 are beautiful and intricate .
for example , in the case of a quintic hypersurface in @xmath58 , the graph depends on choices of triangulations of the two - dimensional faces of the newton polytope of the defining equation ; one such choice is shown on the left side of figure [ fig : triangulation ] .
one constructs the dual graph of each such triangulation , in which each face of the triangulation gives a vertex of the graph , and each edge of the triangulation is crossed by an edge of the graph ( as shown on the right side of figure [ fig : triangulation ] ) . that dual graph then becomes a piece of @xmath45 ( illustrated on the left side of figure [ fig : graphpiece ] )
in which each vertex is `` negative . ''
the pieces are assembled according to the combinatorics of the newton polytope , in which the free ends of the dual graph meet free ends from other faces of the newton polytope , forming trivalent vertices which are the `` positive '' vertices of @xmath45 .
the pieces thus attach three at a time ; a neighborhood of one such attachment is illustrated on the right side of figure [ fig : graphpiece ] for the quintic threefold.,title="fig : " ] for the quintic threefold.,title="fig : " ] a general description of these graphs , for calabi
yau complete intersections in toric varieties , was given by haase and zharkov @xcite and by gross @xcite .
their results show that the classes of mirror pairs described by batyrev @xcite and by batyrev and borisov @xcite admit smooth @xmath0 fibrations satisfying conjecture [ con : topology ] , and that those @xmath0 fibrations are mirror duals of each other . this theory of smooth @xmath0 fibrations of compact calabi
yau threefolds can also be related to some topological aspects of mirror symmetry which have played important rles in the physics literature @xcite .
first , the construction of @xmath45 for a calabi
yau hypersurface @xmath1 in a toric fourfold depends on a choice of triangulation of the faces of the newton polytope , and this choice of triangulation is equivalent to a choice of large complex structure limit point in the moduli space @xcite .
mirror symmetry ( as developed in the physics literature ) offers an alternate interpretation : each large complex structure limit point corresponds to a different birational model of the mirror partner , and the khler cones of the birational models fit together into a common space ( after complexification ) , mirroring the complex structure moduli space @xmath59 @xcite . in this interpretation ,
the choice of birational model depends explicitly on a choice of triangulation of the newton polytope ( which describes the mirror toric fourfold in batyrev s construction @xcite ) .
the simplest birational change a `` flop''is realized by the simple change of triangulation illustrated in figure [ fig : flop ] . as gross pointed out ( * ? ? ?
* remark 4.5 ) , the corresponding change in dual graph ( also illustrated in the figure ) has the correct monodromy properties to be allowed as a new graph @xmath60 .
one expects that by appropriately varying the complex structure and the khler metric , the connecting edge in the original graph will shrink to zero length , giving the @xmath61-valent vertex shown in the intermediate stage ; further variation then causes a new connecting edge to grow , changing the topology to that of @xmath60 . the intermediate step illustrated in the middle of figure [ fig : flop ]
is a partial triangulation , which is expected to correspond to a `` conifold '' singularity on the mirror calabi
yau threefold .
a second topological feature of mirror symmetry is the `` conifold transition '' @xcite in which that conifold singularity is resolved with a small blowup , rather than being smoothed with a change of complex structure .
( the conifold singularity is on the mirror partner , but this transition can also be described on the original calabi yau threefold @xcite . )
the graph @xmath45 appears to change as follows , as proposed independently by gross @xcite and ruan @xcite : after shrinking the connecting edge to zero size , leaving a @xmath61-valent vertex , the two arms of the graph crossing at that vertex are separated into different planes , as illustrated in figure [ fig : conifold ] . as gross and ruan verify , this change is compatible with the monodromies around the edges and produces the expected change in topological euler characteristic for a conifold transition .
however , the relation between this construction and more global versions of the conifold transition remains mysterious and needs further study .
local moduli for a compact special lagrangian submanifold @xmath11 of a compact calabi
yau manifold @xmath1 were determined by mclean @xcite : the deformation space is smooth , and its tangent space is canonically identified with the space of harmonic @xmath27-forms of @xmath11 .
hitchin @xcite used this identification to construct two affine structures on @xmath62 ( if @xmath24 is smooth ) , which geometrize the monodromy transformations that occurred in conjecture [ con : topology ] .
if @xmath63 is a normal vector field to @xmath11 in @xmath1 , then the contraction @xmath64 of @xmath63 with the khler form gives a harmonic @xmath27-form on @xmath11 , and the contraction @xmath65 of @xmath63 with the holomorphic @xmath7-form gives a harmonic @xmath66-form on @xmath11 .
these constructions give rise to the affine structures , which can be seen by considering periods of the harmonic forms .
if @xmath67 , , @xmath68 are vector fields on the deformation space @xmath69 which span the tangent space to @xmath69 at @xmath70 $ ] , and @xmath71 , , @xmath72 is a basis for @xmath73 , then we can form the period matrix @xmath74 the @xmath27-forms @xmath75 on the deformation space @xmath69 are closed , and can be integrated to give local coordinates @xmath76 , , @xmath77 on @xmath69 at @xmath70 $ ] satisfying @xmath78 .
such coordinate systems provide an affine structure , that is , the transition functions between any two such coordinate systems lie in the affine group @xmath79 . intrinsically , the lattice @xmath80 should be identified with @xmath73 in this case , and this affine structure carries the information about the monodromy on @xmath73 .
similarly , if @xmath81 , , @xmath82 is a basis for @xmath83 , then we can form the period matrix @xmath84 the @xmath27-forms @xmath85 on the deformation space @xmath69 of @xmath11 are closed , and can be integrated to give local coordinates @xmath86 , , @xmath87 on @xmath69 satisfying @xmath88 .
such coordinate systems provide the other affine structure , which carries the information about the monodromy on @xmath89 .
hitchin shows that these two affine structures are related by a legendre transform with respect to a suitable locally defined function @xmath90 ( which also determines a canonical metric on the deformation space ) .
this is a version of mirror symmetry which is formulated strictly on the base of the torus fibrations , a notion which was further developed in @xcite .
when @xmath1 is a compact calabi
yau threefold , it is expected that the map @xmath24 giving a supersymmetric torus fibration @xmath91 will be only piecewise smooth , so the analysis of sections [ sec : smooth ] and [ sec : affine ] does not directly apply .
gross showed @xcite that if @xmath24 is smooth , the discriminant locus @xmath92 must have codimension two ; in the more general `` piecewise smooth '' case , @xmath93 may have codimension @xmath27 .
however , we do expect that @xmath93 will always have a retraction onto a subset @xmath45 of codimension two .
and there is a particular limiting situation in which this retraction should become evident : the large complex structure limit . in fact , gross wilson @xcite and kontsevich
soibelman @xcite have formulated a precise conjecture about the large complex structure limit of a supersymmetric torus fibration .
let @xmath94 be a maximally unipotent degeneration of compact simply - connected calabi
yau manifolds of complex dimension @xmath7 , degenerating at @xmath95 , let @xmath96 be a sequence with @xmath97 , and let @xmath98 be a sequence of ricci - flat metrics on @xmath99 with diameter bounded above and below .
then there exists a subsequence @xmath100 which converges in the sense of gromov
hausdorff @xcite to a metric space @xmath101 , where @xmath102 is homeomorphic to to the sphere @xmath103 , and @xmath104 is induced by a riemannian metric on @xmath105 for some @xmath106 of codimension two . following the affine structure to this limit
, one expects to find a limiting affine structure , and in fact the discriminants @xmath107 should have collapsed to @xmath108 in the limit .
harvey and lawson s original paper about calibrations @xcite gave an explicit example of a special lagrangian fibration .
define @xmath109 by @xmath110 then the fibers of @xmath111 are special lagrangian , and are all invariant under the action of the diagonal torus with determinant @xmath27 : @xmath112 the singularities of fibers are located where @xmath113 for some pair of indices @xmath114 and
@xmath115 ; the images of these give three rays within the plane @xmath116 , namely ( i ) @xmath117 , @xmath118 , ( ii ) @xmath119 , @xmath120 , and ( iii ) @xmath121 .
this is illustrated in figure [ fig : trivalent ] .
-plane for the harvey lawson fibration of @xmath122 . ]
the nonsingular fibers are all homeomorphic to @xmath123 .
note that whenever @xmath124 , if we write @xmath125 then either @xmath126 ( in which case @xmath127 ) , or @xmath128 ( in which case @xmath129 ) .
thus , there are two natural subsets @xmath130 of the fiber @xmath131 , distinguished by the sign of @xmath132 .
these subsets meet along @xmath133 when the fiber is smooth , each subset is a manifold with boundary , and they meet along their common boundary .
however , when the fiber is singular and @xmath134 , each subset is itself a smooth special lagrangian submanifold , homeomorphic to @xmath135 . note that @xmath136 is a special case : as harvey and lawson pointed out ,
each subset @xmath137 is a cone over @xmath31 , and those cones meet precisely at the origin in @xmath122 .
joyce @xcite builds some new special lagrangian fibrations by carefully combining subsets of harvey lawson fibers .
let @xmath138 and @xmath139 then @xmath140 and similarly for @xmath141 . to build a special lagrangian fibration , joyce considers translations of these manifolds for @xmath142 .
let @xmath143 these can be made the fibers of special lagrangian fibrations by defining @xmath144 by @xmath145 with this definition , @xmath146 .
notice that the fibrations @xmath147 are only piecewise smooth , and that the discriminant locus in each case is @xmath148 , which has codimension @xmath27 .
notice also that when @xmath149 , both @xmath150 and @xmath151 contain the boundary of the holomorphic disk @xmath152 of area @xmath153 , and that when @xmath154 , both @xmath150 and @xmath151 contain the boundary of the holomorphic disk @xmath155 of area @xmath156 . in some sense , these shrinking disks are `` responsible '' for the singularity being created at @xmath157 .
to go further , joyce invokes the extensive theory which he developed in @xcite concerning the structure of special lagrangian @xmath42-manifolds with a @xmath51 action . using that theory ,
he is able to construct ( * ? ? ?
* theorem 6.5 ) a special lagrangian fibration @xmath158 whose discriminant locus is a _ ribbon _ , that is , the locus @xmath159 , as illustrated in figure [ fig : ribbon ] .
there are several important properties of this example of joyce s .
first , the fiber over an interior point of the ribbon has two singularities
one locally modeled by @xmath160 and the other locally modeled by @xmath161 .
second , as in those local models , there are holomorphic disks with boundary in the fiber for @xmath162 ( and @xmath163 ) , whose area approaches @xmath52 as @xmath164 approaches @xmath52 ; in fact , there is one such holomorphic disk for each of the two singular points .
third , as we approach the boundary of the strip within the plane @xmath124 , something interesting happens : the two bounding circles approach each other and the holomorphic disks cancel out as the boundary of the strip ( either @xmath117 or @xmath165 ) is reached .
there are no holomorphic disks when @xmath166 or @xmath167 .
the region in which holomorphic disks are present is illustrated in figure [ fig : thickribbon ] .
thus , along the boundary of the strip , one has a singularity of multiplicity @xmath5 ( in an appropriate sense ) , which bifurcates into a pair of singularities in the middle of the strip , and those singularities rejoin at the other boundary .
note that the plane which contains the discriminant locus can be identified intrinsically using local affine coordinates .
the cycle @xmath168 which bounds a holomorphic disk is the vanishing cycle for the family , and it defines a dual subspace in @xmath169 , which can be locally identified with the plane containing the discriminant locus
. this plane can also be characterized as the monodromy - invariant plane in the compactly - supported cohomology of the fiber .
joyce conjectures that his examples exhibit generic behavior .
in fact , even lagrangian fibrations ( not just special lagrangian fibrations ) are expected to exhibit these phenomena , as explained in @xcite .
[ con : joyce ] let @xmath32 be a supersymmetric @xmath0 fibration of a compact calabi
yau threefold with respect to a calabi
yau metric whose compatible complex structure is sufficiently close to a large complex structure limit point , and whose khler class is sufficiently deep in the khler cone
. then 1 . the discriminant locus @xmath92 has codimension one . in affine
coordinates , @xmath93 is locally contained in the plane corresponding to the monodromy - invariant subspace of @xmath170 for @xmath171 near @xmath93 .
the fiber over the general point of @xmath93 has two singular points , one of which is modeled locally by @xmath160 and the other of which is modeled locally by @xmath161 .
the fiber over the general point of the boundary of @xmath93 is modeled locally by @xmath172 .
4 . let @xmath173 be the set of fibers which contain the boundary of at least one holomorphic disk in @xmath1 , or are singular .
then the boundary of @xmath93 is contained in the boundary of @xmath174 .
the last statement about which fibers contain the boundaries of disks was not conjectured by joyce , but is consistent with the behavior exhibited by his example @xmath172 ( as illustrated in figure [ fig : thickribbon ] ) .
a common feature of the constructions of zharkov @xcite and ruan @xcite , which dovetails nicely with the analysis of joyce described in section [ sec : noncompact ] , is the description of the discriminant locus of a supersymmetric torus fibration as an _
amoeba_. amoebas were introduced by gelfand , kapranov , and zelevinsky @xcite ; we will briefly review the theory , following mikhalkin @xcite ( see also @xcite ) . .
] let @xmath175 be a laurent polynomial in @xmath7 complex variables .
( here , @xmath176 is a multi - index with negative powers allowed , but @xmath111 has only finitely many non - zero terms . ) the _ amoeba of @xmath111 _ is the set @xmath177 , where @xmath178 , and @xmath179 is defined by @xmath180 a simple example is given by @xmath181 , which is often chosen because it can be graphed exactly , as in figure [ fig : lineamoeba ] .
more complicated examples have `` holes '' in the amoeba , as indicated on the left side of figure [ fig : amoeba ] .
the laurent polynomial @xmath111 has an associated newton polytope @xmath182 and toric variety @xmath183 , and there is a moment map @xmath184 once a symplectic structure has been chosen on @xmath183 . the closure of @xmath185 is called the _ compactified amoeba of @xmath111_. note that @xmath186 and @xmath187 are closely related : the interior of the image @xmath182 of @xmath187 is mapped homeomorphically by @xmath188 to all of @xmath189 .
an example of a compactified amoeba is illustrated on the right side of figure [ fig : amoeba ] .
forsberg , passare , and tsikh @xcite showed that each component of the complement @xmath190 is convex , and that there is an injective map from the set of components to the lattice points @xmath191 in the newton polytope , in which the bounded components ( the `` holes '' ) map to points in the interior of @xmath182 . for a given polyhedron @xmath93 , there exist functions @xmath111 with @xmath192 whose amoebas have the maximum number of holes , but typically there also exist functions with @xmath192 whose amoebas have fewer holes . in the case
@xmath29 , a formula of baker @xcite identifies the genus of a smooth compactification of the affine curve @xmath193 with the number of interior lattice points in the newton polytope .
thus , in that case , the maximum number of holes coincides with the genus .
there is an associated topological picture when the number of holes is maximal : the map @xmath186 will in this case be @xmath5-to-@xmath27 over the interior of the amoeba , and @xmath27-to-@xmath27 on the boundary of the amoeba .
it is easy to see that this gives the right answer for the genus .
additional information about the amoeba can be obtained by considering the _ ronkin function _
@xmath194 defined by @xmath195 ronkin @xcite and passare
rullgrd @xcite show that @xmath196 is well - defined on all of @xmath189 , is convex over @xmath197 , and is locally linear on the complement of @xmath197 .
let @xmath198 be the set of components of the complement , and let @xmath199 be the extension of @xmath200 to a linear function on all of @xmath189 .
passare and rullgrd define @xmath201 which is a piecewise linear function on @xmath189 , and then define the _ spine of the amoeba @xmath197 _ to be the set @xmath202 of points at which the function @xmath203 is not locally linear .
a key theorem of @xcite is that the spine @xmath204 is a strong deformation retract of the amoeba @xmath197 .
note that the spine of the amoeba which is shown in figure [ fig : lineamoeba ] is precisely given by figure [ fig : trivalent ] .
ruan @xcite observed that for a calabi
yau hypersurface @xmath1 in a toric fourfold @xmath205 which is close to the large complex structure limit , the intersections @xmath206 with pairs of toric divisors have amoebas with the maximum number of holes , and these amoebas retract to their spines as the large complex structure limit is approached .
the spines in fact form the pieces of the graph @xmath45 used to describe a topological @xmath0 fibration , which are dual graphs of appropriate triangulations of the newton polytopes ( as was illustrated in figure [ fig : triangulation ] ) .
an amoeba for @xmath207 , together with its spine , is shown in the case of the quintic threefold in figure [ fig : spine ] .
the spine is precisely the graph which occurred on the left side of figure [ fig : graphpiece ] . as in the case of @xmath45 itself , moving among different large complex structure limit points causes the combinatorics of the triangulation to change ( as discussed at the end of section [ sec : smooth ] ) ; we expect a corresponding change in the combinatorics of the amoebas . of the quintic threefold.,title="fig : " ] of the quintic threefold.,title="fig : " ]
gross s reconstruction theorem produced a topological @xmath208-manifold out of the data describing a smooth @xmath0 fibration , and this provides ( in principle ) a method for constructing mirror partners when they are not known , provided that one has a smooth @xmath0 fibration .
castao - bernard and matessi @xcite have proved an analogous theorem which produces a compact symplectic @xmath208-manifold with a piecewise smooth lagrangian @xmath0 fibration , starting from the data describing the smooth fibers of this fibration . the starting data in this case is the affine structure on the base a refinement of the simple monodromy data which gross s theorem needed .
the key technique of stitching together smooth @xmath0 fibrations along a common boundary had been developed in earlier work of these authors @xcite .
a particularly interesting feature of castao - bernard and matessi s construction is the behavior of the discriminant locus @xmath93 .
their fibrations have a discriminant locus which is a trivalent graph near the positive vertices , but has codimension @xmath27 near the negative vertices .
the discriminant locus retracts onto a trivalent graph ; the inverse `` thickening '' of parts of this graph to a codimension @xmath27 set replaces each neighborhood of a negative vertex with an amoeba - like shape which retracts back to the graph .
moreover , the fibration is smooth outside of a set which retracts to ( a subset of ) the graph .
this result , when combined with the discussion in section [ sec : amoebas ] , helps to motivate our conjectures in the next two sections .
prior to joyce s analysis of the structure of special lagrangian fibrations @xcite , there had been speculation that supersymmetric @xmath0 fibrations of compact calabi
yau threefolds would always be smooth , so that the detailed structure ( in the generic case ) would be the one given in conjecture [ con : topology ] .
however , joyce s analysis prompted many of us to rethink the question , and to try to formulate properties analogous to those of conjecture [ con : topology ] which we would expect supersymmetric @xmath0 fibrations to have .
once such formulation appears in conjecture [ con : geometry ] below .
in conjecture [ con : topology ] , the discriminant locus is a graph @xmath45 , but in general we should expect a discriminant locus @xmath93 which only retracts to a graph @xmath45 . as joyce pointed out ( and was already mentioned in conjecture [ con : joyce ] ) , the first thing to expect is that the edges of the graph @xmath45 should thicken to ribbons ; moreover , one should see two singular points in each fiber over an interior point of the ribbon , with the two points coming together to a single singular point along the edges of the ribbon .
the next thing to expect was also proposed by joyce @xcite : since at a `` negative '' vertex , the local monodromy transformations share a common @xmath5-dimensional fixed plane , @xmath93 should remain planar , and the negative vertex should be replaced by a `` trivalent ribbon '' of the sort illustrated in figure [ fig : negvertex ] .
( this is the same structure found by castao - bernard and matessi @xcite in the lagrangian case . )
near a `` positive '' vertex , the three planes containing parts of @xmath93 share a common line but are distinct ; joyce also made a specific proposal for the structure in this case , but we will make a slightly different proposal in our main conjecture below . another motivation for our conjecture is the observation by ruan that in his construction ( and also in zharkov s construction ) , the discriminant locus is built out of amoebas , in fact , out of amoebas with the maximum number of holes . since such amoebas arise from moment maps which are @xmath5-to-@xmath27 over the interior and @xmath27-to-@xmath27 over the edges , it is natural to identify the set of singular points of @xmath24 with the algebraic curve whose moment map image is the amoeba .
this is what we do in our main conjecture .
a third input to our conjecture is the harvey
lawson fibration of @xmath122 , which is the standard model of a `` positive '' vertex with a lagrangian structure . in that fibration ,
the set of singular points consists of the coordinate axes in @xmath122 , which meet in a `` transverse triple point . ''
we conjecture that this is a general property of positive vertices .
[ con : geometry ] let @xmath17 be a supersymmetric @xmath0 fibration of a compact calabi
yau threefold with respect to a calabi
yau metric whose compatible complex structure is sufficiently close to a large complex structure limit point , and whose khler class is sufficiently deep in the khler cone .
then @xmath24 is piecewise smooth and * the set @xmath23 of singular points of fibers of @xmath24 is a complex subvariety of @xmath1 of complex dimension @xmath27 . *
all singular points of @xmath209 are transverse triple points , locally of the form @xmath210 for local complex coordinates @xmath211 , @xmath212 , @xmath213 .
* for each connected component @xmath214 of @xmath209 , @xmath215 is contained in a ( real ) surface @xmath216 , and the map @xmath217 is generically @xmath5-to-@xmath27 onto its image @xmath215 , which has the topology of a compactified amoeba with @xmath218 holes .
* the discriminant locus @xmath93 retracts to a trivalent graph @xmath45 which is the union of the spines of the ( topological ) compactified amoebas @xmath215 .
the graph @xmath45 has all of the properties in conjecture [ con : topology ] . *
the positive vertices of the graph @xmath45 are the points in @xmath93 at which the spines of the various compactified amoebas meet .
the map @xmath24 puts the singular points of @xmath209 in one - to - one correspondence with the positive vertices * the set @xmath219 of fibers which are either singular or contain the boundary of at least one holomorphic disk retracts onto the discriminant locus @xmath93 .
( although in the local example illustrated in figure [ fig : thickribbon ] the set @xmath174 extended far away from @xmath93 , we expect that in global examples @xmath174 will be confined to a small neighborhood of @xmath93 , as illustrated in figure [ fig : cylinder ] . )
the map @xmath24 is smooth to be smooth on the larger set @xmath220 . ] when restricted to @xmath221 .
we have been deliberately vague about the notion of a topological amoeba and its spine , since we do nt know how much of the theory of amoebas should be expected to go through .
it would be very interesting to know , for example , if some version of the ronkin function can be defined for a supersymmetric @xmath0 fibration .
note that one of the things which could happen if we attempt to deform this structure too far away from a large complex structure limit point is that @xmath217 might stop being generically @xmath5-to-@xmath27 , as happens for moment maps of algebraic curves .
it would be very interesting to see what happens to supersymmetric @xmath0 fibrations in that case . presumably , something more general than joyce s phenomenon of two singular points per fiber is going on here ( as joyce briefly discusses in ( * ? ? ?
* section 8.2 ) ) . among the consequences of our conjecture is a specific prediction for the structure of @xmath93 near a positive vertex .
using local affine coordinates to identify a neighborhood of the vertex with the first cohomology of the fiber , the three local pieces of @xmath93 must be contained in the three monodromy - invariant @xmath5-planes , which meet along a common line but are distinct .
however , because the corresponding singular point of @xmath209 is a transverse triple point , the thickening of each piece of the discriminant locus will need to `` thin down '' near the positive vertex so that the three pieces of @xmath93 meet in a single point , leading to a description of the discriminant locus similar to that illustrated in figure [ fig : tripledisc ] .
( this `` thinning down '' was absent from joyce s proposal about the positive vertices . )
notice that , as joyce observed , the discriminant locus has a markedly different local structure near positive and negative vertices and therefore we can not hope for a mirror symmetry statement which simply dualizes all nonsingular tori in the fibration .
the conjecture that @xmath24 is smooth on a region whose complement retracts to @xmath93 was motivated in part by the properties of the construction of castao - bernard and matessi @xcite . the conjecture that @xmath222 provides such a region
is motivated in part by joyce s observation that at least in examples the boundary between the set of tori bounding holomorphic disks and the set not bounding holomorphic disks is a boundary along which @xmath24 fails to be smooth . an additional motivation for part ( f ) of the conjecture is the hope that a proper understanding of the disk contributions to the physical `` t - duality '' construction will restore the symmetry between the @xmath0 fibrations on the original calabi yau manifold and its mirror partner : the fibrations would consist of dual tori on the complement of @xmath222 ( where hitchin s legendre transform will relate the affine structures ) , with the duality between the tori somehow modified within @xmath222 by the disk contributions .
we close with a final conjecture , which is perhaps less well - motivated than conjecture [ con : geometry ] , but which proposes an explanation for why the structures we expect from supersymmetric @xmath0 fibrations are related to the complex structure being near a large complex structure limit point ( as the physics suggests ) .
this final conjecture also points the way towards a connection between our conjectures and the interesting program of gross and siebert @xcite , which formulates mirror symmetry in terms of degenerations of algebraic varieties .
let @xmath223 be a proper flat family of threefolds whose generic point @xmath224 is a calabi
yau threefold , and whose fiber @xmath225 at some special point @xmath226 is a large complex structure degeneration of the form @xmath227 , where the @xmath228 are the components of @xmath225 .
equip @xmath223 with a relative khler metric @xmath229 whose khler classes are sufficiently deep in the khler cone
. then there exist ( non - flat ) families of subvarieties @xmath230 such that @xmath231 , but @xmath232 is nonsingular of complex dimension @xmath27 when @xmath233 is nonsingular , is not equidimensional : the fiber over @xmath234 is a surface while the fiber over a general point @xmath235 is a curve .
] such that for all @xmath235 sufficiently close to @xmath234 there is a supersymmetric @xmath0 fibration of @xmath233 with respect to @xmath236 whose singular locus @xmath237 is precisely @xmath238 .
this is the structure found in the case of calabi yau hypersurfaces in toric fourfolds : in that case , each @xmath239 is an intersection of toric divisors , which meets the nearby nonsingular calabi
yau threefolds @xmath233 in a complex curve @xmath232 ; the union of those curves , in the constructions of zharkov and of ruan , forms the set @xmath237 of singular points of @xmath233 .
it is a pleasure to thank emanuel diaconescu , robbert dijkgraaf , mark gross , christian haase , dominic joyce , ronen plesser , cumrun vafa , and ilia zharkov for useful conversations about the topics reported on here , and to thank the referee for some helpful comments .
m. aganagic , a. klemm , m. mario , and c. vafa , _ the topological vertex _ ,
commun . math .
* 254 * ( 2005 ) 425478 , arxiv : hep - th/0305132
. m. aganagic , a. klemm , and c. vafa , _ disk instantons , mirror symmetry and the duality web _ , z. naturforsch .
a * 57 * ( 2002 ) 128 , arxiv : hep - th/0105045 . m. aganagic and c. vafa , _ mirror symmetry , d - branes and counting holomorphic discs _ , arxiv : hep - th/0012041 .
s. aspinwall , b. r. greene , and d. r. morrison , _ calabi yau moduli space , mirror manifolds and spacetime topology change in string theory _ , nuclear phys .
b * 416 * ( 1994 ) 414480 , arxiv : hep - th/9309097 .
s. aspinwall , c. a. ltken , and g. g. ross , _ construction and couplings of mirror manifolds _ , phys .
b * 241 * ( 1990 ) 373380 .
s. aspinwall and d. r. morrison , _ string theory on k3 surfaces _ , mirror symmetry ii ( b. greene and s .- t .
yau , eds . ) , international press , cambridge , 1997 , pp . 703716 , arxiv : hep - th/9404151 . h. f. baker , _ examples of applications of newton s polygon to the theory of singular points of algebraic functions _
. soc . * 15 * ( 1893 ) 403450 .
v. v. batyrev and l. a. borisov , _ on calabi yau complete intersections in toric varieties _ , higher - dimensional complex varieties ( trento , 1994 ) , de gruyter , berlin , 1996 , pp .
3965 , arxiv : alg - geom/9412017 .
e. calabi , _ on khler manifolds with vanishing canonical class _ , algebraic geometry and topology , a symposium in honor of s. lefschetz ( r. h. fox et al . , eds . ) , princeton university press , princeton , 1957 , pp .
p. candelas , a. m. dale , c. a. ltken , and r. schimmrigk , _ complete intersection calabi
yau manifolds _ , nuclear phys .
b * 298 * ( 1988 ) 493525 .
candelas , x. de la ossa , a. font , s. katz , and d. r. morrison , _ mirror symmetry for two parameter models
, nuclear phys .
b * 416 * ( 1994 ) 481562 , arxiv : hep - th/9308083 .
p. candelas , x. c. de la ossa , p. s. green , and l. parkes , _ a pair of calabi yau manifolds as an exactly soluble superconformal theory _ , nuclear phys .
b * 359 * ( 1991 ) 2174 .
p. candelas , m. lynker , and r. schimmrigk , _ calabi yau manifolds in weighted @xmath240 _ , nuclear phys .
b * 341 * ( 1990 ) 383402 .
r. castao - bernard and d. matessi , _ some piece - wise smooth lagrangian fibrations _ ,
semin . mat .
torino * 63 * ( 2005 ) 223253 .
a. ceresole , r. dauria , s. ferrara , w. lerche , and j. louis , _ picard fuchs equations and special geometry _
phys . a * 8 * ( 1993 ) 79114 , arxiv : hep - th/9204035 . d. a. cox and s. katz , _ mirror symmetry and algebraic geometry _ , mathematical surveys and monographs , vol .
68 , american mathematical society , providence , ri , 1999 .
b. r. greene , d. r. morrison , and a. strominger , _ black hole condensation and the unification of string vacua _ , nuclear phys .
b * 451 * ( 1995 ) 109120 , arxiv : hep - th/9504145 .
b. r. greene and m. r. plesser , _ duality in calabi
yau moduli space _ , nuclear phys .
b * 338 * ( 1990 ) 1537 .
m. gromov , _ structures mtriques pour les varits riemanniennes _ , textes mathmatiques , vol . 1 , cedic , paris , 1981 , edited by j. lafontaine and p. pansu . to3em , _ special lagrangian fibrations .
ii . geometry .
a survey of techniques in the study of special lagrangian fibrations _ , surveys in differential geometry : differential geometry inspired by string theory , surv . differ .
, vol . 5 , international press , cambridge , 1999 , pp .
341403 , arxiv : math.ag/9809072 .
to3em , _ the strominger yau
zaslow conjecture : from torus fibrations to degenerations _ , algebraic geometry
seattle 2005 .
part 1 , proc .
pure math .
80 , amer .
soc . , providence , ri , 2009 , pp .
149192 , arxiv:0802.3407 [ math.ag ] .
s. hosono , a. klemm , s. theisen , and s .- t .
yau , _ mirror symmetry , mirror map and applications to calabi yau hypersurfaces _ , commun .
phys . * 167 * ( 1995 ) 301350 , arxiv : hep - th/9308122 .
d. joyce , _ singularities of special lagrangian fibrations and the syz conjecture _ , comm .
* 11 * ( 2003 ) 859907 , arxiv : math.dg/0011179v3 .
m. kontsevich and y. soibelman , _ homological mirror symmetry and torus fibrations _ , symplectic geometry and mirror symmetry ( seoul , 2000 ) , world sci .
publishing , 2001 , pp . 203263 , arxiv : math.sg/0011041 .
to3em , _ compactifications of moduli spaces inspired by mirror symmetry _ ,
journes de gomtrie algbrique dorsay ( juillet 1992 ) , astrisque , vol .
218 , socit mathmatique de france , 1993 , pp . 243271 , arxiv : alg - geom/9304007 .
to3em , _ beyond the khler cone _ , proc . of the hirzebruch 65 conference on algebraic geometry ( m. teicher , ed . ) , israel math .
conf . proc .
, vol . 9 , bar - ilan university , 1996 , pp . 361376 , arxiv : alg - geom/9407007 .
to3em , _ the geometry underlying mirror symmetry _ , new trends in algebraic geometry ( k. hulek , f. catanese , c. peters , and m. reid , eds . ) , london math .
lecture notes , vol .
264 , cambridge university press , 1999 , pp .
283310 , arxiv : alg - geom/9608006 .
to3em , _ through the looking glass _ , mirror symmetry iii ( d. h. phong , l. vinet , and s .- t .
yau , eds . ) , ams / ip studies in advanced mathematics , vol . 10 , american mathematical society and international press , 1999 , pp . 263277 , arxiv : alg - geom/9705028 .
to3em , _ geometric aspects of mirror symmetry _ , mathematics unlimited 2001 and beyond ( b. enquist and w. schmid , eds . ) , springer - verlag , 2001 , pp . 899918 , arxiv : math.ag/0007090 . m. passare and h. rullgrd , _ amoebas , monge - ampre measures , and triangulations of the newton polytope _ , duke math .
j. * 121 * ( 2004 ) 481507 .
ruan , _ lagrangian torus fibration of quintic hypersurfaces . i. fermat quintic case _
, winter school on mirror symmetry , vector bundles and lagrangian submanifolds ( cambridge , ma , 1999 ) , ams / ip stud . adv . math .
, vol . 23 , amer . math
, 2001 , pp . 297332 , arxiv : math.dg/9904012 .
to3em , _ lagrangian torus fibration of quintic calabi
yau hypersurfaces .
symplectic topological syz mirror construction for general quintics _ , j. differential geom .
* 63 * ( 2003 ) 171229 , arxiv : math.dg/9909126 .
to3em , _ lagrangian torus fibrations and mirror symmetry of calabi yau manifolds _ , symplectic geometry and mirror symmetry ( seoul , 2000 ) , world sci .
publishing , 2001 , pp . 385427 , arxiv : math.dg/0104010 .
to3em , _ generalized special lagrangian torus fibration for calabi
yau hypersurfaces in toric varieties ii _ , mirror symmetry v ( n. yui , s .-
yau , and j. d. lewis , eds . ) , american mathematical society and international press , 2007 , arxiv : math.dg/0303278 .
a. strominger , s .-
yau , and e. zaslow , _ mirror symmetry is t - duality _ , nuclear phys .
b * 479 * ( 1996 ) 243259 , arxiv : hep - th/9606040 .
e. witten , _ phases of @xmath243 theories in two dimensions _ , nuclear phys .
b * 403 * ( 1993 ) 159222 , arxiv : hep - th/9301042 . s .-
yau , _ calabi s conjecture and some new results in algebraic geometry _ , proc
* 74 * ( 1977 ) 17981799 . | we formulate some precise conjectures concerning the existence and structure of supersymmetric @xmath0 fibrations of calabi yau threefolds , and describe how these conjectural fibrations would give rise to the strominger yau
zaslow version of mirror symmetry .
mirror symmetry between calabi
yau manifolds remains , some twenty years after its discovery , one of the biggest mysteries in mathematics .
originally formulated as a physical relationship between certain pairs of calabi yau manifolds @xcite ( with astonishing mathematical consequences relating enumerative geometry to hodge theory @xcite ) , the mirror symmetry proposal was refined in 1996 by strominger , yau , and zaslow @xcite into a much more geometric statement , again based on physics .
but while the main idea of the strominger
yau
zaslow proposal has been clear from the outset , many of the details have remained elusive .
one of the reasons for this is that the proposal involves special lagrangian submanifolds of a calabi
yau manifold , and very few tools are available for studying such submanifolds . as a consequence , the initial period of intense study of the original strominger yau
zaslow proposal has largely ended , and much of the recent work on mirror symmetry has shifted to other approaches ( including reformulations of the strominger yau
zaslow proposal ) , as is recounted in detail elsewhere in this volume .
our purpose in this paper is to give a quite precise conjectural formulation of the strominger
yau zaslow version of mirror symmetry for calabi yau threefolds .
the conjectures we formulate are modifications of conjectures previously made by gross , ruan , and joyce ; we also restate some other conjectures from @xcite .
our formulation is unfortunately not directly based on examples , since as mentioned above
tools for constructing concrete examples are not currently available .
the conjectures are , however , motivated on the one hand by qualitative features of the strominger yau
zaslow proposal which have been discovered by mathematicians , and on the other hand by some suggestive arguments from physics .
we focus on the motivation from mathematics in this paper . for simplicity ,
we restrict our attention for the most part to calabi yau manifolds of complex dimension one , two , or three .
we expect , however , that similar conjectures could be formulated in higher dimension , at the expense of greater combinatorial complexity . |
now that there is good evidence from reactor @xcite and accelerator experiments @xcite supporting the solar matter oscillation @xcite and atmospheric oscillation solutions , @xcite the lsnd @xcite puzzle is the remaining oscillation mystery , and soon miniboone will release new evidence which should resolve that issue .
the solar and atmospheric data are consistent with a minimal three - family oscillation , with each described by a mass - squared difference and a mixing angle . the parameter space of the remaining mixing angle has been explored by experiments , including reactor neutrino experiments , and this mixing angle is constrained to be small .
evidence from solar , atmospheric , and reactor experiments is so compelling , that it is now common to refer to the `` standard model '' which includes at least two non - zero neutrino masses .
lsnd does not fit within this three - angle , two mass - squared difference picture , and many authors @xcite have proposed alternatives . here
we consider an alternative interpretation of the data if miniboone s observations are not consistent with the standard model , and if non - standard neutrino interactions are of significant size .
neutrinos for miniboone are created when 8 gev protons collide with a beryllium target producing 3 gev secondary mesons which decay into mostly muon neutrinos with 0.6 @xmath3 electron neutrinos expected .
the neutrino energy is around 0.5 gev , the neutrinos travel 541 m to the detector , and they re detected primarily through quasi - elastic scattering , @xmath4 .
there are significant backgrounds , and the miniboone collaboration has studied related experiments and planned carefully for overcoming these , including significant improvement in understanding of intrinsic @xmath0 from muon and kaon decays , neutral pions with only one photon ring observed , the small @xmath5 rate , and how to distinguish @xmath0 from @xmath2.@xcite our question is whether non - standard neutrino interactions could be responsible if more than the expected @xmath0 are observed , and what would happen if non - standard interactions resulted in @xmath1 .
direct limits on wrong - flavor neutrinos in pion decay are not as strong as one might expect , and limits through @xmath6 symmetry from all - charged - lepton modes may be avoided in some models .
the direct limit on @xmath7 is 0.008 and no direct limit on @xmath8 exists ; indirect limits through @xmath6 symmetry are much stronger.@xcite several authors have previously studied non - standard interactions in various neutrino contexts @xcite since lepton - flavor violating interactions are present in many models that have neutrino mass and mixing .
our model includes neutrino oscillations and a low - energy effective four - fermi interaction that may violate lepton flavor number in a unified way .
@xcite these kinds of flavor - violating interactions may occur in a number of models , and here we remain as model - independent as possible . in this paper
we report on a three - flavor analysis of what miniboone may expect to observe if the conventional solar and atmospheric solutions are correct and there are flavor - changing direct interactions .
we have previously discussed a low - energy , effective four - fermi model for including non - standard neutrino interactions with neutrino oscillations and studying their consequences . here
we simplify the notation used in our previous papers by omitting the subscript corresponding to the flavor of the charged lepton ; we assume that a meson has decayed to a muon and a neutrino , and so only a ( created ) neutrino flavor subscript is required .
@xmath9^\dag + h.c.\ ] ] in this paper we consider only standard helicity and lorentz structure ( encapsulated in @xmath10 and @xmath11 ) and the semi - leptonic interactions relevant to miniboone .
the subscripts @xmath12 and @xmath13 are allowed to be @xmath14 and @xmath15 , respectively , for the neutrino flavor and mass eigenstates .
we use the conventional three - flavor mixing matrix , @xmath16 , as a unitary transformation between these flavor and mass eigenstates . including neutrino oscillations and these non - standard interactions at the source ,
the flavor - change probability is : @xmath17 here @xmath12 and @xmath18 are summed over all three flavors with @xmath19 and @xmath20 for the flavor expected in the standard model .
we parameterize the new interaction strengths with a magnitude given by a pseudo - angle , @xmath21 , and a cp phase , @xmath22 , and there is a pair of these for each of the possible unexpected neutrino flavors , @xmath23 and @xmath24 .
notice , then , that the detected wrong - flavor neutrino may have had several different origins : 1 ) the correct flavor ( @xmath2 ) was produced but the neutrino oscillated to @xmath0 , 2 ) the wrong flavor ( @xmath0 ) was produced in the pion decay , and 3 ) the wrong flavor ( @xmath1 ) was produced in the pion decay and oscillated to the detected wrong flavor ( @xmath0 ) .
miniboone employs a large number of pmts looking into a spherical signal region and a small number of pmts looking outward into a veto region in order to identify cerenkov cones created after an electron or muon neutrino interacts in the detector , creating a charged @xmath23 or @xmath25 .
the collaboration has studied how to distinguish between them ( and discriminate from background ) using multiple algorithm particle identification .
the experiment is a @xmath0 appearance experiment with high sensitivity , but if the simplest three - neutrino mixing model is correct , miniboone will not detect @xmath0 above background .
the experiment was devised in order to collect enough data to persuasively confirm or refute lsnd s intriguing observations roughly a decade ago . for the bulk of our study , we considered the case of @xmath26 and @xmath27 .
the effects of all the cp phases ( including delta from the standard model ) were negligible , at most a fraction of a percent .
we used values of @xmath28 and @xmath29 consistent with solar , atmospheric , reactor , and accelerator ( except lsnd ) data . @xcite
as shown in figure [ fig : figure1 ] , direct interactions are unlikely to produce enough @xmath0 to be observable by miniboone ; a flavor - change probability on the order of @xmath30 would correspond to a handful of @xmath0 among the couple thousand expected . even if all of this handful were observed , they would be among the hundreds of events either containing or misidentified as @xmath0 , a signal - to - background of less than 0.01 .
this also means miniboone is unlikely to be able to improve the limits on @xmath31 . that leaves the possibility that @xmath1 arrive in the detector .
figure [ fig : figure1 ] also shows that the constraints on production of tau neutrinos at the source are much weaker .
obviously @xmath24 ca nt be produced through quasi - elastic scattering of neutrinos with such low energy ; however , if @xmath1 are present in the beam , then the flux of @xmath2 observed would be lower by a corresponding amount .
given that the miniboone collaboration will know the number of protons on target to about 1 percent , they may be able to use muon disappearance to observe non - standard interactions or set a direct limit on @xmath32 . if miniboone can perform the @xmath2 disappearance experiment , then figure [ fig : figure2 ] can be used to approximate either @xmath32 or their new limit on it .
we have completed a three - flavor oscillation plus non - standard interaction analysis of miniboone .
we showed that while the types of new interactions we studied here are unlikely to result in an observable @xmath0 appearance signal , they might lead to a measurable deficit of @xmath2 .
we thank doug mckay and mary hall reno for discussions .
this work was supported in part by the kamerling fund .
we mourn the death of our collaborator , williams .
figure [ fig : figure1 ] displays the low probability of a @xmath0 being produced with these non - standard interactions in addition to oscillations using parameters consistent with solar , atmospheric , and reactor data .
also shown is the much higher flavor - change probability for @xmath1 ; however , these will be challenging to detect since @xmath24 ca nt be produced from such a low - energy beam .
both flavor - change probabilities are relatively independent of @xmath33 .
figure [ fig : figure2 ] displays how rapidly the flavor - change probability climbs as the direct non - standard interaction strength , @xmath32 , increases .
since the @xmath1 probability is approximately flat over the energy range for miniboone in figure [ fig : figure1 ] , this second figure can be used to see what values of @xmath32 the experiment is sensitive to ; if one percent too few @xmath2 are observed and the number of @xmath0 corresponds only to the expected background , then @xmath32 is around 0.1 . | miniboone is expected to soon report analysis of their closed data set , and we here consider what effect direct non - standard neutrino interactions would have on their observations .
current direct limits on non - standard interactions make interpretation of a @xmath0 signal at miniboone as three - flavor oscillations plus direct interactions untenable .
however , @xmath1 from non - standard interactions at the source could contribute to a @xmath2 deficit which may be observable if non - standard interactions are sizable .
neutrino , oscillation , neutrino , flavor , pi+ , leptonic decay , interaction , neutrino , numerical calculations , interpretation of experiments 13.15.+g , 14.60.pq , 12.60.-i |
since the report of superconductivity in mgcni@xmath5 @xcite , antiperovskite compounds of type m@xmath5ax ( m = mn , fe , ni ; a = ga , al , sn , zn , cu , in , ge , ag ; x = c , n ) have been extensively investigated @xcite .
interest in mn based materials of this type in recent times has been due to properties like giant magnetoresistance ( gmr ) @xcite , negative thermal expansion ( nte ) @xcite and near zero temperature coefficients of the resistivity ( nz - tcr ) @xcite .
the wide range in physical properties suggests possibilities of a wider range of chemical bonding via partial replacement of constituent elements that could control the magnetostructural correlations in these materials @xcite . among these , mn@xmath6gac has attracted considerable attention due to its multiple magnetic transitions .
with decreasing temperature it exhibits three phase transitions : first , a second - order transition from paramagnetic ( pm ) to a collinear ferromagnetic ( fm ) phase at curie temperature ( t@xmath7 ) @xmath8 .
the next transition is to a canted ferromagnetic ( cfm ) phase at t @xmath9 via a second - order transition @xcite and the third is a first - order transition to the antiferromaagnetic ( afm ) ground state at @xmath10 @xcite accompanied by a discontinuous change in volume @xcite and a 5k thermal - hysteresis @xcite .
contrastingly , the compound mn@xmath11snc presents a spontaneous magnetization change from the room temperature pm state to a ferrimagnetic ( fim ) one via a first - order transition around t@xmath12 @xcite .
results of neutron diffraction suggest that the compound possesses a complicated spin arrangement consisting of afm and fm sub - lattices perpendicular to each other resulting in a non - collinear fim state @xcite .
unlike its ga containing counterpart , mn@xmath0snc undergoes only one magnetic transition from pm to fim state which also involves an abrupt change in its unit cell volume @xcite .
various attempts to understand properties accompanying the first - order transition have been made by altering the external conditions such as pressure and dopants @xcite .
effect of pressure on mn@xmath6gac is to stabilize the fm phase as observed through increase of t@xmath13 and a concomitant decrease in first - order transformation temperature @xcite .
further , high pressure x - ray diffraction studies up to 35 gpa revealed no structural transition @xcite .
transition metal substitutions at the mn sites result in weakening the first - order transformation .
fe doping in mn@xmath6snc not only gives rise to a competition between the afm and fm components in the fim state resulting in a broadened fm region but also quenches the nte behavior reported in mn@xmath5snc @xcite .
ni or co doping in mn@xmath0gac results in a gradual decrease in the afm first - order transition temperature @xcite . in ni - doped compounds ,
the mr exceeds 60%-70% at 5 t which is attributed to the weakened afm ground state .
variation of carbon content has the most dramatic effects .
increasing the concentration of carbon in mn@xmath11snc@xmath14 leads to a growth of fm sub - lattice at the expense of afm sub - lattice @xcite while the carbon deficiency in the compound mn@xmath11gac@xmath15 results in a total disappearance of the first - order afm ground state at low temperatures @xcite .
strong correlations between the metalloid ( c , n ) and its octahedron of surrounding mn atoms allows the substitution of various metals at the a - site @xcite .
ge - doping in mn@xmath5x@xmath16ge@xmath17c ( x = al , zn , ga , sn ) causes local lattice distortions which in turn provokes a local magnetic disorder reducing the first - order magnetic transition temperature .
ge doping in mn@xmath5snc gradually reduces the mve , till it disappears .
substitution of al for ga in mn@xmath5gac produces the same effect as the application of high pressure while partial substitution of zn increases t@xmath18 and maximal values of mr along with gradually suppressing the afm ground state . from the above it
is clear that doping suppresses or changes the character of first - order magnetic transformation indicating the importance of correlations between mn and other constituents . mn@xmath0gac and mn@xmath0snc are two compounds which exhibit a contrasting evolution of their properties . while mn@xmath0gac has an afm ground state , its sn counterpart orders ferrimagnetically . with different doping fm state is seen to stabilize in mn@xmath0gac , afm interactions strengthen in mn@xmath0snc .
there is no clear understanding of transformation from an antiferromagnetic ground state to a ferrimagnetic one as ga is replaced by sn .
a critical balance of ferromagnetic and antiferromagnetic interactions is required for the materials applicability in magnetic refrigeration .
while ferromagnetic ordering is responsible for large changes in magnetic entropy , antiferromagnetic order suppresses hysteresis
. therefore it would be interesting to study the evolution of the first - order magnetic transformation in the solid solutions of the type mn@xmath0ga@xmath1sn@xmath2c , @xmath19 .
in order to synthesize polycrystalline samples of general formula mn@xmath5ga@xmath20sn@xmath15c , ( x=0 , 0.1 , 0.2 , 0.4 , 0.5 , 0.6 , 0.8 , 0.9 & 1.0 ) , desired proportions of the starting materials were weighed , mixed and pressed into pellets before annealing them in evacuated quartz tubes at 1073k for the first 48 hours and 1150k for 120 hours @xcite .
about 15 wt.% excess carbon was added to realize in stoichiometric carbon compounds @xcite . after cooling to room temperature the phase formation and purity of the samples
were examined by recording x - ray diffraction ( xrd ) patterns using cu - k@xmath21 radiation .
studies of the magnetic properties of the samples were carried out in a quantum design magnetic property measurement system ( mpms ) in the temperature range 5k to 390k and in applied magnetic fields of @xmath22 and @xmath23 and between @xmath24 .
magnetization as a function of temperature was recorded under three experimental modes : zero - field cooling ( zfc ) , field - cooled - cooling ( fcc ) , and field - cooled warming ( fcw ) in an applied magnetic field of 0.01 t . for the zfc mode , each sample was cooled down to 5k from room temperature in the remanant field ( @xmath25 0.001 t ) before applying the necessary magnetic field .
the zfc data was then recorded while warming .
the subsequent cooling and warming magnetization curves with applied field on , are termed as fcc and fcw respectively .
isothermal magnetization at different temperatures in the field range of @xmath26 were recorded in zfc mode .
similarly electrical resistivity of the samples was measured by standard four probe method using a quantum design physical properties measurement system ( ppms ) in the temperature range 5k and 300k during cooling and warming cycles .
x - ray diffraction patterns recorded in the @xmath27 range were rietveld refined using fullprof suite .
refinement revealed that the xrd patterns could be well indexed using the structural model ( space group : _ pm@xmath28 m _ ) reported for mn@xmath5gac as shown in figure [ fig : xrd ] along with minor impurity phases of mno ( @xmath29 ) and sno ( @xmath30 ) sn(@xmath31 ) and graphite ( maximum @xmath32 ) .
along with global parameters like , scale factor and background , lattice parameter and sn / ga ratio were refined for the major phase .
the refined concentrations of sn at ga site are listed in table [ tbl : riet1 ] .
it can be seen that compounds sn1 and sn2 have similar values of sn concentration and so is the case with sn6 and sn7 which could be due to volatility of ga .
ga@xmath20sn@xmath15c , @xmath19 refined using rietveld method . ]
.values of sn concentration obtained from rietveld refinement of x - ray diffraction patterns . [ cols="^,^,^",options="header " , ] a plot of lattice constants as a function of sn concentration in figure [ fig : lattice ] with values obtained from refinement shows a systematic increase across the entire composition range .
this result is consistent with the fact that sn atoms are larger than those of ga and successfully replace them at the a - site .
such a replacement by sn results in an increased cell volume with increase in doping concentration of sn @xcite thus obeying vegard s law . as a function
sn concentration in mn@xmath5ga@xmath20sn@xmath15c , @xmath19 . ] electrical resistivity measurements for the mn - based antiperovskites show a metallic behaviour over the entire temperature range .
figure [ fig : res1 ] compares temperature dependent resistivity plots for the mn@xmath5ga@xmath20sn@xmath15c series of compounds measured during the cooling and warming cycles .
temperature dependence of resistivity for all the samples is quite similar .
figure [ fig : res1]a shows the variation of resistivity in sn2 compound .
a distinct change in slope is observed at about @xmath33 followed by a minimum and a sharp rise thereafter at t @xmath34 . in case of mn@xmath0gac ,
these temperatures indicate second and first - order magnetic transitions respectively @xcite . the strong discontinuity and hysteresis observed between cooling and warming curves all along
the series suggests that the compounds in the series undergo a first - order type of transition . with increase in sn concentration
although overall metallic behavior remains the same , the minimum indicating first - order transition peak exhibits a gradual shift towards higher temperatures .
another change observed in resistivity curves is decrease in sharpness of the resistivity increase after the minimum in case of sn3 and sn4 compounds .
the sharpness of this transition returns for compounds with higher sn concentration . in the case of the other end member
the series , sn8 ( mn@xmath0snc ) , the anomaly in resistivity at about t @xmath35 , coincides with pm to fim transformation @xcite .
ga@xmath36sn@xmath2c , @xmath19 ] in order to understand the nature of both first - order and second - order transformations in mn@xmath5ga@xmath20sn@xmath15 magnetization measurements were carried out .
the results of these measurements allow us to classify the compounds in the following three categories : 1 .
gallium rich compounds ( sn0 , sn1 & sn2 ) 2 . intermediate compounds ( sn3 , sn4 & sn5 ) each of which will be discussed separately 3 . tin rich compounds ( sn6 , sn7 & sn8 ) figure [ fig : mtga ] depicts magnetization curves recorded in zfc , fcc and fcw modes in applied field of @xmath22 .
magnetization as a function of temperature m(t ) of mn@xmath37gac ( sn0 ) exhibits a second - order transition from the high temperature pm phase to a fm one at t@xmath38 = @xmath39 followed by a first - order transformation to afm state at t@xmath40 @xmath41 .
as expected , the m(t ) plots for the samples sn1 and sn2 are similar to those of sn0 , exhibiting a pm - fm transition at t@xmath12 @xmath42 and t@xmath12 @xmath43 respectively .
the first - order transformations to afm ground state are also seen at t@xmath40 @xmath44 and t@xmath45 @xmath46 respectively for sn1 and sn2 samples .
these are lower than the first - order transformation temperature of sn0 indicating sn doping in mn@xmath0gac atleast in ga rich compounds decreases both the first - order and second - order transformation temperatures .
in order to understand the effect of sn in greater detail , isothermal magnetization curves were recorded and studied at selected temperatures in the @xmath47 field range .
a comparison of hysteresis loops recorded at @xmath48 for sn0 and sn2 are shown in figure [ fig : mtga]d .
although both the compounds are afm at @xmath48 , m(h ) loops exhibit a curvature but with non saturating magnetic moment even in a field of @xmath49 .
the magnetization increases noticeably with sn replacing ga in the sn2 compound .
another distinct feature of the hysteresis loop is the initial magnetization curve lying outside the loop .
this suggest the existence fm interactions in the compound even at @xmath48 . to further understand
the magnetic interactions in sn doped compound , the isothermal magnetic response at select temperatures were recorded and are plotted in figure [ fig : mhsn1 ] . while the @xmath50 loop depicts paramagnetic behaviour , the loops at @xmath51 and @xmath52 indicate ferromagnetic ordering of the sample .
afm ground state of the sample is clear from the hysteresis loop recorded at @xmath48 with interfering fm interactions as shown earlier .
the m(h ) loops recorded just below the afm transformation , show a clear metamagnetic transition from afm to fm state .
it can also be seen that the closer one is to the transformation temperature , the afm - fm transition occurs in lower applied fields . such a field induced transition has been also seen in mn@xmath0gac near its transformation temperature indicating that the magnetic interactions in these low sn concentration samples are similar to those in the undoped mn@xmath0gac .
effect of sn doping at these lower concentrations seems to introduce ferromagnetic interactions which can be clearly seen in m(h ) loop recorded at t = @xmath48 .
it also lowers the first - order fm - afm transformation temperature .
here we discuss magnetization measurements on three intermediate compositions sn3 , sn4 and sn5 individually .
temperature dependent magnetization of sn3 presented in figure [ fig : sn3]a reveals an interesting picture .
contrary to expectations that increasing sn doping would support ferromagnetic order at the expense of antiferromagnetism , the ferromagnetic ordering temperature of sn3 compound decreases while the antiferromagnetic transformation temperature increases as compared to sn2 compound .
furthermore , it appears that even before the fm transition is complete the compound abruptly transforms to the afm state via a first - order transition at t @xmath9 .
however , the hysteresis loops in figure [ fig : sn3]b . and c. recorded at specific temperatures depict that although the compound has dominant afm interactions , fm correlations are also present . the hysteresis loops recorded at all temperature
@xmath53 @xmath51 have a non saturating @xmath54s shaped curvature indicating presence of competing fm and afm interactions .
the hysteresis loop recorded at t = @xmath52 , which is just below its transformation temperature exhibits a weak metamagnetic transition . at @xmath48 the virgin magnetization curve lies outside the envelope hinting at presence of strongly competing fm interactions . from the hysteresis loops
, it appears that with decreasing temperature , both fm and afm interactions strengthen . at about 50% doping of sn for ga ,
the magnetization results are even more interesting .
the m(t ) curves recorded in zfc , fcc and fcw modes for sn4 compound are displayed in figure [ fig : sn4 ] .
magnetization recorded in a applied magnetic field of @xmath22 exhibits a relatively sharp upturn indicating some kind of ferromagnetic ordering at about t @xmath55 .
absence of hysteresis between the fcc and fcw magnetization curves indicates the transition to be of second - order .
however , as in sn3 even before the transformation to ferromagnetic state is complete , magnetization drops sharply only to rise again even more strongly below @xmath56 .
both these transitions appear to be first - order in nature as indicated by presence of hysteresis between fcc and fcw magnetization curves .
it appears as if one part of the sample tends to favour antiferromagnetic alignment of the mn spins during structural deformation , while the other part facilitates fm order .
one could also argue that the compound is essentially ferromagnetic along with magnetic inhomogeneities .
if this be the case , then higher magnetic field should favor ferromagnetic alignment and smear out the signatures of antiferromagnetic ordering seen at around @xmath52 .
however , both fm and afm transitions are present in magnetization recorded at h = @xmath23 and @xmath57 . furthermore ,
both these transformation temperatures reduce with increasing magnetic field . another confirmation for presence of fm and afm regions
is obtained from isothermal magnetization recorded at select temperatures .
presence of a hysteresis loop at @xmath48 indicates the compound to be ferromagnetic as also indicated by m(t ) curves . however , the non saturation magnetic moment up to h = @xmath49 is indicative of presence of antiferromagnetic interactions . at higher temperatures ( @xmath58 , @xmath59 ) , the m(h ) loops indicate weakening ferromagnetism and a gradual built up of antiferromagnetic interactions . at @xmath60 , the m(h ) loop is of a typical antiferromagnet but with a metamagnetic transition .
such metamagnetic transitions are seen in all ga rich samples and are typical of mn@xmath0gac .
these observations lend weight to the argument that with increasing ga concentration , afm regions grow at the expense of fm regions .
results of m(t ) measurements carried out on sn5 compound ( figure [ fig : sn5 ] ) show that as the concentration of tin is further increased to about @xmath6170% , the compound transforms from a pm state to a fm like ordering via a first - order transition at t@xmath38 @xmath62
. however , there are other characteristics that also dominate the magnetization behavior .
firstly , there is a large divergence between zfc and fc curves indicating presence of fm and afm regions . secondly , the magnetization peaks at about @xmath60 and then passes through a minimum at about @xmath63 .
presence of hysteresis between fcc and fcw magnetization curves in this entire temperature range indicates that the alignment of mn spins is continuously evolving through out the first - order transformation .
it is known that mn@xmath0snc consist of fm and afm sublattices at low temperatures @xcite .
decrease of carbon content in mn@xmath0snc increases afm regions at the expense of fm regions . in sn5 compound
, it appears that the presence of nearly 30% ga forces mn spins to also align antiferromagnetically .
this is further accentuated by m(h ) curves .
hysteresis loop recorded at @xmath48 indicates dominance of fm interaction however , the magnetic moment does not saturate even in magnetic fields up to @xmath49 .
the m(h ) curves recorded in the range of @xmath58 @xmath3 t @xmath3 @xmath51 exhibit similar characteristics .
the value of magnetization increases with decreasing temperature except below @xmath59 where a reversal of trend is noticed .
while the magnetization values recorded at @xmath58 and @xmath59 are similar , those at @xmath48 are distinctly smaller .
this confirms the presence of strongly competing ferro and antiferromagnetic regions . ]
results of magnetization measurements carried out on tin rich compounds sn6 , sn7 & sn8 are shown in figure [ fig : mtsn ] and figure [ fig : mhsn6 ] .
mn@xmath0snc ( sn8 ) has a ferrimagnetic ground state and exhibits a single first - order transformation from paramagnetic state at about @xmath35 .
with increasing ga doping , this transformation temperature decreases to about @xmath64 .
another noticeable feature is the increasing difference in magnetization recorded in zfc and fc modes with increasing ga doping .
this difference tends to diminish in magnetization curves recorded in higher magnetic fields and hence it can be attributed to magnetic inhomogeneities introduced by ga doping .
magnetic hysteresis loops recorded at @xmath48 for sn7 and sn8 are presented in figure [ fig : mtsn]d .
the two loops are expectedly very similar to each other as the high fields wash out the effect of magnetic inhomogeneities .
for both samples , magnetization increases steeply and tends to saturate above @xmath23 demonstrating a typical soft ferromagnetic character .
field - dependent magnetization curves at select temperatures above and below t@xmath38 = @xmath65 for sn6 compound are plotted in figure [ fig : mhsn6 ] .
m(h ) loop recorded at @xmath50 and @xmath64 are in paramagnetic state . while those recorded below @xmath66 confirm the presence of ferromagnetic ordering .
ga@xmath20sn@xmath15c , @xmath19 system ( afm= antiferromagnetic phase , fm = ferromagnetic phase fim = ferrimagnetic phase and pm = paramagnetic phase ) . b. variation of net magnetic moment per formula unit with increasing concentrations of sn . ] in mn@xmath0ga@xmath1sn@xmath2c compounds , while the ga rich compounds have an antiferromagnetic ground state , the sn rich compounds exhibit dominance of ferromagnetic interactions over antiferromagnetic ones . in the intermediate range , with increasing ga concentration ,
a growth of afm regions at the expense of fm regions can be seen from magnetization measurements .
this is also supported by the variation of various transition temperatures and the magnetic moment values obtained from m(h ) curves at @xmath48 .
the transition temperatures of the first - and second - order magnetic transformations for the mn@xmath5ga@xmath20sn@xmath15c compounds were determined from the first derivative of the zfc curves . as illustrated in figure [ fig : phase]a , all the ga rich samples exhibit a second order paramagnetic to ferromagnetic transition before undergoing a magneto - structural transformation to an antiferromagnetic state .
the @xmath66 of this second order transition decreases with increasing sn concentration before disappearing or merging into the first - order transformation curve .
the first - order transformation for ga rich samples is from a ferromagnetic state to an antiferromagnetic state . again with decreasing ga content ,
this transformation temperature too decreases , though very gradually and again rises quite steeply in the sn rich region ( @xmath67 ) .
however , for these compounds , the transformation is from a paramagnetic state to a ferrimagnetic state as reported in case of mn@xmath0snc .
this indicates that the ga rich samples , ( @xmath68 ) basically have antiferromagnetic ground state but with presence of ferromagnetic interactions .
these ferromagnetic interactions could be field induced as in case of mn@xmath0gac or could be due to presence of sn .
the sn rich samples on the other hand ( @xmath67 ) have ferrimangetic ground state which is a mixture of ferromagnetic and antiferromagnetic interactions with antiferromagnetic interactions dominating with increasing ga doping .
this is again clear from a plot of net magnetic moment per formula unit estimated from m(h ) curves recorded at @xmath48 .
this net magnetic moment in each compound was obtained by linear extrapolation of the magnetization value at h = 7 t to meet the magnetization axis at h = 0 t .
these values plotted in figure [ fig : phase]b .
it can be seen that ga rich compounds have much lower magnetic moment values as compared to sn rich compounds .
the magnetic moment values also show a sudden jump around @xmath69 , which seems to indicate a change in magnetic order .
however , these intermediate range compounds ( @xmath70 ) exhibit multiple magnetic transitions and are discussed separately . in mn@xmath0ga@xmath71sn@xmath72c
, it can be seen that there are two magnetostructural transformation .
it seems as if there exist two magnetic phases which are ordering independently of each other .
it is plausible that the regions that is rich in ga order antiferromagnetically while those rich in sn order ferrimagnetically .
the existence of two magnetic phases can be also seen in compounds with @xmath73 = 0.4 and @xmath73 = 0.7 . in the first case ( @xmath73 = 0.4 )
, the ga rich phase first tends to order ferromagnetically via a second order transition and then orders antiferromagnetically through a magnetostructural transformation .
this transformation temperature is lower than that in mn@xmath0gac or 20% sn doped sample .
however , even in the antiferromagnetic phase , there are ample signatures of existence of ferromagnetic correlations .
a comparison of m(h ) loops recorded at 5k in sn0 , sn2 and sn4 shows a systematic increase in magnetization and coercive field with increasing sn concentration .
these are signatures of stronger ferromagnetism .
the situation is completely reversed in mn@xmath0ga@xmath74sn@xmath75c . here
the compound first orders ferromagnetically via a first - order transformation as in case of mn@xmath0snc or 20% ga doped compound but at lower temperatures , the magnetization suddenly dips giving indication of an antiferromagnetic transformation .
the presence of hysteresis between magnetization values recorded while warming and cooling indicates the transformation to be of first - order in nature . again
the m(h ) loops recorded at lower temperatures show a decrease in values of magnetization below 50k .
the initial magnetization curves recorded at 50k and 5k are seen to be lying outside the hysteresis loop .
these are again clear signatures of presence of two competing magnetic phases .
therefore the magnetization studies carried out on mn@xmath0ga@xmath36sn@xmath2c type antiperovskites show a evolution of first - order magnetic transformation from ferromagnetic - antiferromagnetic type in ga rich compounds to paramagnetic - ferrimagnetic type in sn rich compounds . in the intermediate doping range ,
the magnetization studies indicate co - existence of two magnetic phases : ga rich phase which orders antiferromagnetically and a sn rich phase with dominant ferromagnetic interactions .
in conclusion , magnetization of mn@xmath0ga@xmath36sn@xmath2c , @xmath19 series of compounds indicates that though sn doping does not alter the cubic structure or metallic behavior of the compounds it gradually increases the strength of ferromagnetic interactions across the series while suppressing the antiferromagnetic ground state .
the ga - rich compounds ( @xmath76 ) transform to an antiferromagnetic ground state via a first - order transition but with a decreasing first order transition temperature with sn doping .
a similar behavior is noticed in case of sn rich compounds ( @xmath77 ) wherein the para to ferrimagnetic ordering temperature decreases as ga is doped in mn@xmath0snc . in the intermediate concentration range ( @xmath70 ) ,
a co - existence of ferromagnetic and antiferromagnetic order is seen in their magnetic properties .
the results indicate that as sn is doped at the ga - site in mn@xmath0ga@xmath36sn@xmath2c , ferromagnetic sub - lattice grows at the expense of antiferromagnetic sub - lattice to a point where the first - order transition is altered from a fm - afm type in ga rich compounds to a pm - fim type in mn@xmath0snc .
authors thank board of research in nuclear sciences ( brns ) for the financial support under the project 2011/37p/06 .
m / s devendra d. buddhikot and ganesh jangam are acknowledged for the experimental assistance .
99 he t. , huang q. , ramirez a. p. , wang y. , regan k. a. , rogado n. , hayward m. a. , haas m. k. , slusky j. s. , inumara k. , zandbergen h. w. , ong n. p. and cava r. j. 2001 nature * 411 * , 54 li y. b. , li w. f. , feng w. j. , zhang y. q. and zhang z. d. 2005 phys .
b * 72 * , 024411 tong p. , sun y. p. , zhu x. b. and song w. h. 2006 phys
b * 73 * , 245106 tong p. , sun y. p. , zhu x. b. and song w. h. 2006 phys
b * 74 * , 224416 tong p. , sun y. p. , zhao b. c. , zhu x. b. and song w. h. 2006 solid state commun . *
138 * , 64 wang b. s. , tong p. , sun y. p. , li l. j. , tang w. , lu w. j. , zhu x. b. , yang z. r. and song w. h. 2009 appl .
. lett . * 95 * , 222509 wang b. s. , tong p. , sun y. p. et al .
2009 j. appl
. phys . * 105 * , 083907 wang b. s. , tong p. , sun y. p. , zhu x. b. , yang z. r. et al .
2010 appl .
lett . * 97 * , 042508 wang b. s. , lin j. c. , tong p.
2010 j. appl .
phys . * 108 * , 093925 takenaka k. and takagi h. 2005 appl .
87 * , 261902 takenaka k. , asano k. , misawa m. and takagi h. 2008 appl .
. lett . * 92 * , 011927 chi e. o. , kim w. s. and hu n. h. 2001 solid state commun . * 120 * , 307 lin s. , wang b. s. , lin j. c. , et al .
2011 appl .
. lett . * 99 * , 172503 kamishima k. , goto t. , nakagawa h. , miura n. , ohashi m. , mori n. , sasaki t. and kanomata t. 2001 phys .
b * 63 * , 024426 iikubo s. , kodama k. , takenaka k. , takagi h. , takigawa m. and shamoto s. 2008 phys . rev
* 101 * , no .
20 , article i 01 sun y. , wang c. , wen y. c. , zhu k. g. and zhao j. t. 2007 appl .
. lett . * 91 * , 231913 sun y. , wang c. , chu l. , wen y. , m. nie , and liu f. 2010 scr .
mater . * 62 * , 686 takenaka k. and takagi h. 2005 appl .
lett . * 87 * , 261902 kamishima k. , bartashevich m. i. , goto t. , kikuchi m. and kanomata t. 1998 j. phys . soc
. jpn . * 67 * , 1748 kamishima k. , goto t. , sasaki t. , kanomata t. and inami t. 2002 j. phys .
. jpn . * 71 * , 922 bouchaud j. p. , fruchart r. , guillot m. , bartholin h. and chaisse f. 1965 c. r. acad .
paris * 261 * , 655 cakir o. and acet m. 2012 appl .
phys . lett . *
100 * , 202404 fruchart d. , bertaut e.f . and
le clerc b. , veillet p. , lorthioir g. , mme fruchart e. and fruchart r. 1973 j. solid state chem . * 8 * 182 fruchart d. and bertaut e.f .
1978 j. phys . soc
. jpn . * 44 * , 781 wang b. s. , tong p. , sun y. p. , luo x. , zhu x. b. , li g. , zhu x. d. , zhang s. b. , yang z. r. , song w. h. and j. m. dai 2009 epl * 85 * , 47004 wen y. , wang c. , sun y. , nie m. , fang l. and tian y. 2009 solid state commun . * 149 * , 1519 wang b.s .
, tong p. , et al .
2009 j. appl
. phys . * 106 * , 013906 kaneko t. , kanomata t. and shirakawa k. 1987 j. phys .
. jpn . * 56 * , 4047 meenakshi s. , vijayakumar v. , lausi a. and busetto e. 2006 solid state commun .
* 140 * , 500 nie m. , wanga c. , wena y. , suna y. , naa y. , chua l. and tangb m. 2011 solid state commun . * 151 * , 377 wang b.s . , tong p. , sun y.p .
, tang w. , li l.j .
, zhu x.b .
, yang z.r . and song w.h .
2010 physica b,*405 * , 2427 tohei t. , wada h. and kanomata t. 2004 j. magn .
magn . mater .
* e585 * , 272 wen y. , wang c. , nie m. , sun y. , chu l. and dong c. 2010 appl .
* 96 * , 041903 bouchaud j.p . and fruchart r. 1965 c. r. acad .
sci . * 261 * , 458 lewis l. h. , yoder d. , moodenbaugh a. r. , fischer d. a. and yu m - h .
2006 j. phys . : condens .
matter * 18 * 16771686 sun y. , wang c. , na y. , chu l. , wen y. and nie m. 2010 mater .
bull . * 45 * , 1230 yu m - h , lewis l.h . and
moodenbaugh a.r .
2003 j. appl
. phy . * 93 * , 10128 - 10130 suna y. , wanga c. , chua l. , wena y. , naa y. , nie m. and chenb x. 2012 solid state commun .
* 152 * , 446 | the magnetic properties of the mn@xmath0ga@xmath1sn@xmath2c , 0 @xmath3 x @xmath3 1 antiperovskite compounds have been investigated in detail . though all compounds of this series crystallize in a cubic structure , the ga rich ( x @xmath3 0.2 ) compounds transform , via a first order transformation , to an antiferromagnetic ground state and the sn rich ( x @xmath4 0.8 ) compounds exhibit dominant ferromagnetic interactions at low temperature . in the intermediate range ( 0.4 @xmath3 x @xmath3 0.7 ) co - existence of ferromagnetic and antiferromagnetic interactions can be seen .
the results have been explained to be due to growth of ferromagnetic sub - lattice at the expense of antiferromagnetic sub - lattice with increasing sn concentration in mn@xmath0ga@xmath1sn@xmath2c .
this growth occurs to a point where the first - order transition is altered from a ferromagnetic - antiferromagnetic type in ga rich compounds to a paramagnetic - ferrimagnetic type in mn@xmath0snc . |
the discovery of high temperature superconductivity in oxypnictide phases , prototype lafeas(o , f ) by kamihara and co - workers @xcite has sparked widespread interest in establishing the physical properties of these materials and especially the mechanism for superconductivity .
this interest has led to the discovery of a number of new phases including other oxy - arsenides that when electron doped have critical temperatures exceeding 55k , @xcite and hole doped superconducting phases including ( k , ba)fe@xmath0as@xmath0 , with @xmath1=38k @xcite and li@xmath2feas with @xmath1 of 18k .
@xcite thus there are three families discovered so far : ( 1 ) the oxy - arsenides , spacegroup @xmath3 , with rare earth oxide layers separating the feas layers common to all of these superconductors , ( 2 ) the body centered tetragonal , @xmath4 thcr@xmath0si@xmath0 structure materials ( bafe@xmath0as@xmath0 ) , where the feas layers are stacked so that the as atoms face each other and the ba atoms sit in the resulting 8-fold coordinated square prismatic sites between them , and ( 3 ) @xmath3 lifeas , which is like the oxy - arsenide structure but with the rare earth oxide layers removed and replaced by li ( see below ) , resulting in a large decrease in the @xmath5-axis spacing of the feas layers .
stoichiometric ( undoped ) bafe@xmath0as@xmath0 shows a spin density wave ( sdw ) type magnetic order and a structural transition with similar ordering temperatures .
this feature is common to most of these materials , including the oxy - arsenides .
@xcite however , there are also some strong differences between properties of the oxypnictides and bafe@xmath0as@xmath0 .
most notable is the fact that with the exception of one report , @xcite the oxy - arsenide superconductors are generally electron doped , while ( ba , k)fe@xmath0as@xmath0 and li@xmath2feas are found to superconduct when hole doped .
however , in both bafe@xmath0as@xmath0 and lafeaso superconductivity apparently is associated with the suppression of the sdw .
this is potentially of importance because there has been much discussion of the association between magnetic quantum critical points and high temperature superconductivity and in addition the specific spin fluctuations associated with the nesting that presumably drives the sdw have been studied as a possible pairing mechanism .
@xcite electronic structure studies , @xcite show that the fermi surfaces and band structures of these phases are qualitatively rather similar . in particular ,
all the compounds show small compensating electron and hole fermi surfaces , but high densities of states .
the hole fermi surfaces occur around the zone center , and are generally derived from heavier ( lower velocity ) bands than the electron surfaces , which are around the zone corner ( @xmath6 point in a primitive tetragonal zone ) . within band structure theories ,
moment formation is governed by a stoner parameter , @xmath7 , which takes values of 0.7 ev to 0.9 ev for ions near the middle of the @xmath8 series ( note that the effective @xmath7 can reduced by hybridization ) .
while magnetism may occur with lower values of the density of states , it must occur within a band picture if the stoner criterion , @xmath9 , is met ( the non - spin - polarized electronic structure becomes unstable against ferromagnetism in this case ) . here
@xmath10 is the density of states ( dos ) at the fermi energy on a per atom per spin basis .
the calculated values of @xmath10 for the undoped feas materials put them at the borderline of itinerant magnetism , and in fact they are near both ferromagnetism , as required by the stoner theory , and checkerboard antiferromagnetism , where nearest neighbor fe atoms are oppositely polarized .
the ground states , both experimentally @xcite and theoretically @xcite have a magnetic structure that corresponds to condensation of an @xmath6 point ( zone corner ) sdw . the electronic structures all show a strong increase in the electronic dos as the fermi energy is lowered into the heavier hole bands .
this feature is quite robust in this family and is also the case for the corresponding phosphide compounds .
@xcite importantly , this strong increase in the dos poses a conundrum because it means that within a rigid band picture hole doped materials would be strongly magnetic , and therefore would necessarily have very different properties , including superconductivity , from the electron doped materials . on the other hand , from an experimental point of view the generic phase diagram is much more symmetric , with an sdw and structural distortion near zero doping , with apparently non - magnetic , and superconducting states for both hole and electron doped materials .
here we investigate this issue using calculations for doped bafe@xmath0as@xmath0 and lifeas within the virtual crystal approximation .
this is an average potential approximation , which is beyond the rigid band method .
we in fact find strong non - rigid band effects that leave only a weak doping level dependence to the @xmath10 and provide a plausible explanation for the generic phase diagram of this family of materials .
the present calculations were performed within the local density approximation ( lda ) using the general potential linearized augmented planewave ( lapw ) method , @xcite similar to those reported previously for lafeaso .
@xcite lapw sphere radii of 1.8 @xmath11 , 2.2 @xmath11 , 2.1 @xmath11 , and 2.1 @xmath11 were used for li , ba , fe and as , respectively . the effect of doping was included by the virtual crystal approximation , by varying the nuclear charge of ba or li .
importantly , the internal parameters , including the as and li positions , were relaxed independently for each doping level .
we took the experimental tetragonal lattice parameters of room temperature bafe@xmath0as@xmath0 of ref . , and relaxed the internal coordinate of as using lda total energy minimization .
the calculated coordinate is @xmath12=0.342 .
this is noticeably lower than the reported value of @xmath12=0.3545 .
in fact the as heights in these two structures differ by 0.16 , which as discussed below is outside the normal range of density functional errors and is significant for the electronic structures .
we also did calculations with the generalized gradient approximation but find a similar discrepancy .
similar discrepancies have also been noted in the lafeaso series .
@xcite for lifeas , we used the experimental lattice parameters@xcite and relaxed the free internal coordinates .
the experimental structure ( spacegroup @xmath3 , 129 ) shows two li sites : 2@xmath5 , which according to ref .
is fully occupied , and lies above and below the centers of the fe squares opposite the as , as shown in fig .
[ struct ] , and 2@xmath13 which has a reported occupancy of 0.1 in li@xmath14feas and lies directly above the fe .
we performed total energy calculations for stoichiometric lifeas with li in the 2@xmath13 and 2@xmath5 sites and found that the @xmath15 site is favored by 0.52 ev per formula unit .
accordingly , we performed our doping studies placing li in the @xmath15 site .
we note that while this is the opposite choice to that made by nekrasov and co - workers @xcite , the electronic structure near the fermi energy is practically unaffected by this choice .
in fact , when hole doped by li vacancies one may expect coulomb repulsions between the like charged li ions to favor partial occupations of the @xmath16 site , though always with a higher occupancy on the @xmath15 site , similar to what is found in na@xmath17coo@xmath0 . in any case , the insensitivity of the electronic structure to the li position supports the validity of the virtual crystal approximation .
the structures that we used for the undoped compounds are summarized in table [ tab - struct ] .
( color online ) crystal structure of bafe@xmath0as@xmath0 ( left ) and lifeas ( right ) with the relaxed as and li coordinates .
note that in both structures the feas layers are uniformly spaced along the @xmath5-axis direction .
, width=307 ] .structures used in calculations for undoped bafe@xmath0as@xmath0 and lifeas . [ cols="<,^,^",options="header " , ]
the calculated lda band structures and electronic dos for bafe@xmath0as@xmath0 are given in figs .
[ ba - bands ] and [ ba - dos ] , respectively .
these were calculated using the lda relaxed as height as given in table [ tab - struct ] .
the fermi surfaces for the calculated as height are depicted in figs .
[ ba - fs ] and [ li - fs ] .
as noted previously , @xcite they consist of electron cylinders at the zone center and hole cylinders and other sections that depend strongly on the as height at the zone center .
for bafe@xmath0as@xmath0 the hole fermi surface flares out at @xmath18=1/2 ( @xmath19 ) , giving it a more three dimensional character than that of the li compound .
this flaring out is sensitive to the as height , and in particular is reduced as the as is moved away from the fe plane . in the li compound
there are three hole sections . these are two very two dimensional hole cylinders and an inner surface that forms a capped cylinder centered at @xmath20 .
calculations were also done for the experimental as height ( fig .
[ ba - dos ] ) .
these lead to a significantly higher @xmath10 , which will make the material more magnetic , as has been discussed for the oxy - arsenide materials .
@xcite the band structure and dos for lifeas are given in figs .
[ li - bands ] and [ li - dos ] , respectively , again using the calculated internal coordinates .
the calculated values of @xmath10 are 3.06 ev@xmath21 per formula unit ( two fe atoms ) both spins for bafe@xmath0as@xmath0 and 3.58 ev@xmath21 for lifeas on a per unit cell ( two fe atoms , two formula units ) , both spins basis . for the experimental @xmath12=0.3545 of bafe@xmath0as@xmath0
we obtain @xmath10 = 4.59 ev@xmath21 per formula unit .
these values are lower than the corresponding value for lafeaso ( 2.62 ev@xmath21 per fe both spins ) . @xcite but are still high enough to place the materials near magnetism .
calculated lda band structure of bafe@xmath0as@xmath0 for the lda value of @xmath12 .
the lower panel shows a blow - up near @xmath22 , which is set to 0 ev .
the short @xmath20-@xmath19 direction is from [ 0,0,0 ] to [ 0,0,1/2 ] while the long @xmath20-@xmath19 is in the basal plane and runs from [ 0,0,0 ] to [ 1,0,1/2 ] in units of the body centered tetragonal primitive reciprocal lattice vectors . , title="fig:",height=316 ] calculated lda band structure of bafe@xmath0as@xmath0 for the lda value of @xmath12 .
the lower panel shows a blow - up near @xmath22 , which is set to 0 ev .
the short @xmath20-@xmath19 direction is from [ 0,0,0 ] to [ 0,0,1/2 ] while the long @xmath20-@xmath19 is in the basal plane and runs from [ 0,0,0 ] to [ 1,0,1/2 ] in units of the body centered tetragonal primitive reciprocal lattice vectors . , title="fig:",height=316 ] ( color online ) calculated electron dos of bafe@xmath0as@xmath0 for the experimental ( top ) and lda ( bottom ) values of @xmath12 , plotted on a per formula unit basis .
the projections are onto the lapw spheres . since as has extended @xmath23 orbitals that extend well beyond the lapw sphere , this yields values that are proportional to the as contribution , but underestimate it . , width=316 ] calculated lda band structure of lifeas for the lda internal coordinates .
, height=316 ] ( color online ) calculated electron dos of lifeas for the lda internal coordinates , on a per formula unit basis .
note that the formula unit for this compound contains one fe , while that for bafe@xmath0as@xmath0 contains two .
, height=316 ] ( color online ) lda fermi surface of bafe@xmath0as@xmath0 for the lda internal coordinates , shaded by band velocity.,width=316 ] ( color online ) lda fermi surface of lifeas for the lda internal coordinates , shaded by band velocity.,width=316 ] for bafe@xmath0as@xmath0 we find an antiferromagnetic state corresponding to the sdw to be the most stable of the states considered . for the lda value of @xmath12
we do not find any instability against either a ferromagnetic or a checkerboard ( nearest neighbor ) antiferromagnetic state in the lda .
the sdw state , which consists of lines of parallel spin fe atoms in the feas planes has a moment defined by the integral inside the fe lapw sphere , radius 2.1 @xmath11 , of 0.7 @xmath24 .
as mentioned , @xmath10 is larger when the as height is raised to the reported experimental value leading to a more magnetic state . in this case , within the lda we find a very weak instability of the non - spin - polarized state to ferromagnetism ( 0.3 @xmath24/fe , 0.01 mev / fe ) . a stronger instability for the checkerboard antiferromagnetism ( 1.6 @xmath24/fe , 41 mev / fe ) and the strongest instability to the sdw , which is the ground state ( 1.75 @xmath24/fe , 92 mev / fe ) . for the experimental value of @xmath12 the state with ferromagnetic @xmath5-axis stacking of the fe in the sdw state has higher energy , and is therefore less stable than the state with antiferromagnetic stacking , by 3 mev / fe , while these two states are degenerate to the precision of the calculation for the lda as height .
thus the lowest energy magnetic structure that we find is the sdw ordering in the fe layers , stacked so that fe atoms directly above each other in the @xmath5-direction are antiferromagnetically aligned .
this is in accord with recent experimental results .
@xcite the high sensitivity of the fe moment to the ordering and the fact that the sdw state is so much lower in energy than any other magnetic state strongly imply that the magnetism is of itinerant character with a spin density wave driven by the structure of the fermi surface rather than local moment physics associated with heisenberg type exchange couplings .
these undoped electronic structures are similar to those obtained previously and show the generic features of the calculated electronic structures of the other feas materials small fermi surfaces , with hole cylinders at the zone center and electron cylinders at the zone corner , high @xmath10 , and a strongly increasing dos below @xmath22 .
the somewhat lower values of @xmath10 compared to lafeaso would place bafe@xmath0as@xmath0 and lifeas further from ferromagnetism .
this combined with the fact that ferromagnetic spin fluctuations are highly pair breaking for singlet superconducting states may partially explain why hole doping is effective in these non - oxide materials , but the large increase in the dos below @xmath22 works against this explanation taken by itself .
in this section we report virtual crystal calculations for the two compounds .
this approach differs from rigid band in that it includes the self - consistent rearrangement of the charge density , which is done in an average potential .
this approximation is justified since the ba and li states do not contribute in any significant way to the band structure near @xmath22 and because the scattering due to disorder on the ba and li sites is expected to be weak in view of the fact that wavefunctions of the states near @xmath22 are primarily fe derived and therefore separated from the ba / li atoms .
this is supported by the fact that , as mentioned , we find no noticeable difference in the electronic structure as calculated with the li atoms of lifeas in sites @xmath16 and @xmath15 .
importantly , in the present context , we are able to relax the internal coordinates and in particular the as heights in the virtual crystal approximation
. calculated as height as a function of virtual crystal doping , @xmath25 in carriers per fe .
negative @xmath25 corresponds to hole doping.,height=316 ] the calculated as height above the fe plane as a function of doping is shown in fig .
[ zas ] . as may be seen , the as drops towards the fe plane as holes are introduced .
this is as might be anticipated for a primarily ionic situation , since a more highly positively charged fe layer will be more attractive to as@xmath26 and since the effective size of transition element ions in solids decreases as the valence increases .
@xcite the essential point is that in these materials the carrier density is relatively low with an electronic structure near @xmath22 formed from band edges . as a result
the dos and fermi surface are sensitive to small changes , including the as position .
rigid band dos as compared to the virtual crystal @xmath10 as a function of doping level for bafe@xmath0as@xmath0 .
note that the doping levels given in carriers per fe with negative indicating hole doping , and the @xmath10 are given on a per fe atom basis.,height=316 ] rigid band dos as compared to the virtual crystal @xmath10 as a function of doping level for lifeas .
note that the doping levels given in carriers per fe with negative indicating hole doping , and the @xmath10 are given on a per fe atom basis.,height=316 ] figs .
[ ba - rbdos ] and [ li - rbdos ] show the calculated virtual crystal values of @xmath10 for various doping levels as a function of doping level .
as may be seen the behavior of the virtual crystal is very different from what may be expected from a rigid band viewpoint . in particular
, @xmath10 is found to be very nearly constant over the doping range studied , while as mentioned , the rigid band point of view would lead one to expect an increasing @xmath10 upon hole doping .
our result differs from that of a recent calculation of shein and ivanovskii , @xcite who studied a supercell of composition ba@xmath27k@xmath27fe@xmath0as@xmath0 , and found a significant increase in @xmath10 relative to the undoped compound .
we ascribe this difference to the fact that they used the fixed experimental as position for the undoped compound and did not relax the internal coordinates . in any case
, we find that contrary to the rigid band picture hole doping does not bring the system closer to stoner ferromagnetism , which as mentioned would be detrimental to superconductivity . thus the main effects on the electronic structure are more subtle .
( color online ) virtual crystal fermi surface of bafe@xmath0as@xmath0 as a function of doping level @xmath25 using the relaxed as position for each value of @xmath25 .
the shading is by velocity .
the left panels show a view along the @xmath18 directions , while the right panels are along @xmath28 . , width=316 ] ( color online ) virtual crystal fermi surface of lifeas as a function of doping level @xmath25 using the relaxed atomic positions for each value of @xmath25 .
the shading is by velocity .
the left panels show a view along the @xmath18 directions , while the right panels are along @xmath28 .
the outermost hole cylinder is very two dimensional and therefore is difficult to see in the view along @xmath18 .
accordingly we indicate its position by an arrow . , width=316 ] figs .
[ ba - fsdope ] and [ li - fsdope ] show the evolution of the non - spin - polarized virtual crystal fermi surface for the two compounds .
as may be seen the basic structure of the fermi surface with hole sections around the zone center and electron sections at the zone corner is maintained for all the doping levels shown .
the difference between the volume of the hole and electron sections is required to correspond to the doping level by the luttinger theorem .
also it may be seen that the size of the electron sections changes at least as much as that of the hole sections ( note that in fig .
[ ba - fsdope ] the top view emphasizes the flaring out of the hole fermi surface at @xmath18=0.5 , while the main cylindrical sections are smaller ) .
this is contrary to what might be anticipated considering that the electron bands are more dispersive than the hole bands and therefore should change less and again shows the importance of non - rigid band effects .
the condition for an sdw instability of the paramagnetic state is that the real part of the susceptibility , @xmath29 should diverge at the nesting vector .
this function is given in terms of an integral over the fermi surface for the bare lindhard susceptibility , @xmath30 , with an enhancement factor that emphasizes peaks , much like the stoner enhancement of the pauli susceptibility for the ferromagnetic case .
the interband lindhard susceptibility for an electron cylinder at the zone corner and a same sized hole cylinder at the zone center is a function peaked at the nesting vector , while if the cylinders differ in radius by @xmath31 , e.g. due to doping or extra hole sections , the function will have a flat cylindrical plateau centered at the former nesting vector with diameter @xmath32 .
thus as the cylinders become mismatched in size the maximum value of @xmath30 ( the value at the nesting vector ) will decrease , while a region of high constant @xmath30 will develop within @xmath31 of the nesting vector .
we calculated the lindhard function , setting all matrix elements to be equal ( i.e. constant matrix element approximation yielding arbitrary units ) .
although the fermi surfaces are not perfect cylinders and there are multiple surfaces of different size , we do in fact find this behavior for lifeas .
it is also the case for electron doping in bafe@xmath0as@xmath0 and to a lesser degree for hole doped bafe@xmath0as@xmath0 , where the peak in @xmath30 becomes broader and smaller but there is no clear plateau region . for both lifeas and bafe@xmath0as@xmath0 we find a decrease in the peak height of the lindhard function as we dope with holes .
this explains why the sdw is destroyed by hole doping in these materials .
there has been much recent discussion of the relationship between the sdw instability and superconductivity . within an eliashberg formalism involving spin fluctuations what enters
the calculation of the pairing function is a double integral over the fermi surface ( like that in the lindhard function ) with a kernel containing the pairing interaction , @xmath33 .
thus in the case discussed above , the interband superconducting pairing between the electron and hole fermi surfaces will depend on an integral over a range of @xmath34 of size comparable to the fermi surface size , while the sdw instability will depend on the peak value of @xmath29 ( in a real case spin fluctuations away from the sdw ordering vector will compete with the sdw so even for a given maximum value of the bare @xmath30 a system with a narrower peak will order first ) . in any case , if spin fluctuations associated with the sdw are responsible for pairing , one expects superconductivity to compete with the sdw and to occur on both sides of the region where the sdw is stable .
what our results show is that the properties are much more symmetric between hole and electron doping than would be expected from a rigid band picture .
y. kamihara , t. watanabe , m. hirano , and h. hosono , j. am . chem . soc .
* 130 * , 3296 ( 2008 ) .
ren , w. lu , j. yang , w. yi , x.l .
shen , z.c .
che , x.l .
dong , l.l .
sun , f. zhou , and z.x .
zhao , chin .
lett . * 25 * , 2215 ( 2008 ) . | we report density functional calculations of the electronic structure and fermi surface of the bafe@xmath0as@xmath0 and lifeas phases including doping via the virtual crystal approximation .
the results show that contrary to a rigid band picture , the density of states at the fermi energy is only weakly doping dependent and that the main effect of doping is a change in the relative sizes of the electron and hole fermi surfaces as required by luttinger s theory .
this is a consequence of a change in as height with doping , in particular a shift of as towards fe as holes are introduced in the fe plane , as might be expected from simple ionic considerations .
the main effect of doping is therefore a reduction in the degree of nesting of the fermi surface .
this provides a framework for understanding the approximate electron - hole symmetry in the phase diagrams of the fe - as based superconductors . |
modern observations like high redshift survey of sne ia @xcite , cmbr @xcite , wmap @xcite illustrate the accelerating phase of the universe . accelerating phase phenomenon of the universe
is not understood by the standard big bang model of cosmology giving a time like singularity in the past .
cosmological constant @xmath13 with the equation of state @xmath14 is the simplest candidate in einstein s gravity which describe the present accelerating phase of the universe . the aspects of inflation and the cosmological constant are not understood well .
there are various candidates to play the role of the dark energy having negative pressure and describe the accenting view of the present observation .
the mystifying fluid namely dark energy is understood to dominate the 70% of the universe which violets the strong energy condition and 30% dark matter ( cold dark matters plus baryons ) .
various effective candidate of dark energy namely chaplygin gas with equation of state @xmath15 , @xmath16 introduced by @xcite .
furthermore it has been generalized to the form @xmath17 , @xmath18 @xcite and modified to @xmath19 @xcite . the another candidate of dark energy was introduced by chakraborty et al @xcite , known as _ new variable modified chaplygin gas _ ( nvmcg ) which follows the equation @xmath20 , @xmath21 , @xmath22 being the scale factor which gives interesting physical significance .
+ for flat universe having the energy densities for dust like matter and dark energies we need to know the value of critical energy density and @xmath23 at high accuracy @xcite .
the mcg best fits with the 3 year wmap and the sdss data with the choice of parameters @xmath24 and @xmath25 @xcite which are improved constraints than the previous ones @xmath26 @xcite .
here we have assumed the new variable modified chaplygin gas model in flat frw cosmology .
the joint data analysis of stern data set with bao and cmb have been analyzed for this model .
the distance modulus vs redshift has been examined for our model via redshift - magnitude observational data from supernova type ia ( union 2 ) .
the friedmann - robertson - walker(frw ) metric is considered as @xmath27\end{aligned}\ ] ] where @xmath28 is the scale factor of the universe and @xmath29 , @xmath30 , @xmath31 are the dimensionless comoving co - ordinates , @xmath32 be the curvature parameter of space - time metric and takes the values @xmath33 for flat , closed , open universe respectively . + we consider a spatially flat universe ( @xmath34 ) with dark energy and dark matter ( non - interacting ) .
thus the einstein equations becomes ( choosing @xmath35 ) @xmath36 @xmath37 where @xmath38 the total energy density of the universe and @xmath39 the total pressure .
@xmath40 , @xmath41 are the energy density and pressure for dark energy respectively and @xmath42 , @xmath43 that for dark matter . for non - interacting fluid conservation equations
become @xmath44 and @xmath45 we have assumed that the universe is filled with new variable modified chaplygin gas ( nvmcg ) as dark energy whose eos is @xcite @xmath46 where @xmath47 , and @xmath48 are function of the scale factor @xmath22 . in particular , choose @xmath49 and @xmath50 with @xmath3 , @xmath4 , @xmath6 , @xmath7 are positive constants . for @xmath51 ,
this model reduces to modified chaplygin gas and for @xmath52 , the model reduces to the variable modified chaplygin gas model . + expression for the energy density for nvmcg model
is obtained from ( 4 ) as @xcite @xmath53^{\frac{1}{1+\alpha}}\end{aligned}\ ] ] where @xmath5 is an integration constant and @xmath54 is the upper incomplete gamma function . + for dark matter , the eos is @xmath55 and so @xmath56 , @xmath57 .
the hubble parameter ( @xmath0 ) in terms of redshift parameter ( @xmath9 ) can be expressed as ( from eq .
( 2 ) ) @xmath58 @xmath59\end{aligned}\ ] ] subsequently , we investigate the bound on the model parameter by observational data fitting .
the parameters are determined by @xmath23-@xmath9 ( stern ) , bao and cmb data analysis @xcite using @xmath11 minimization technique from hubble - redshift data set .
from the above expression we can write , the hubble parameter @xmath23 which can be put in the form as @xmath60 where , @xmath61 @xmath62^{1/2}\end{aligned}\ ] ] now @xmath63 contains six unknown parameters @xmath64 and @xmath8 .
now we will fixing two parameters and by observational data set the relation between the other two parameters will obtain and find the bounds of the parameters . [
cols="<",options="header " , ] table 1 : the hubble parameter @xmath23 and the standard error @xmath65 for different values of redshift z. for given @xmath66 , @xmath3 and @xmath4 can be best fitted by minimizing @xmath67 given by @xmath68 where @xmath69 is the observed hubble parameter at redshift @xmath9 and @xmath70 is the error associated with that particular observation ( see table 1 ) and @xmath0 represents the theoretical values of hubble parameter calculated for our model .
here we use the observed value of hubble parameter at different redshifts ( twelve data points ) listed in observed hubble data by stern et al @xcite to analyze our model .
we consider the present value of hubble parameter @xmath71 and a fixed prior distribution . by fixing the model parameter
@xmath72 $ ] , @xmath73 and @xmath5 , we determine the range of other two parameters @xmath3 and @xmath4 by minimizing ( 11 ) .
the probability distribution function in terms of the parameters @xmath3 , @xmath4 , @xmath5 , @xmath73 and @xmath8 is given by @xmath74 where @xmath75 is the prior distribution function for @xmath1 .
now our best fit analysis with stern observational data support the theoretical range of the parameters . in figures 1 and 2 , we plot the graphs for different confidence levels 66% ( solid , blue ) , 90% ( dashed , red ) and 99% ( dashed , black ) contours for @xmath76 and @xmath77 respectively and by fixing the other parameters .
the best fit values of @xmath78 , @xmath79 and the minimum values of @xmath80 are tabulated in table 2 .
@xmath81 * table 2 : * @xmath23-@xmath9 ( stern ) : the best fit values of @xmath78 , @xmath79 and the minimum values of @xmath80 for @xmath82 , @xmath83 , @xmath84 and for different values of @xmath8 .
= 3 in = 2.5 in = 3 in = 2.5 in fig.1 fig.2 figs .
1 and 2 show the variation of @xmath3 with @xmath4 for @xmath76 and @xmath85 respectively for different confidence levels .
the 66% ( solid , blue ) , 90% ( dashed , red ) and 99% ( dashed , black ) contours are plotted for the @xmath23-@xmath9 ( stern ) analysis . in this section we use the method of joint analysis , the baryon acoustic oscillation ( bao ) peak parameter proposed by eisenstein et al @xcite .
acoustic peaks occurred because cosmological perturbations excite sound waves in initial relativistic plasma in the early epoch of the universe .
sloan digital sky survey ( sdss ) survey is one of the first redshift survey ( 46748 luminous red galaxies spectroscopic sample ) by which the bao signal has been directly detected at a scale @xmath86 100 mpc .
the corresponding comoving scale of the sound horizon shell is about 150 mpc in radius .
we shall investigate the two parameters @xmath3 and @xmath4 for our model using the bao peak joint analysis for low redshift ( with range @xmath87 ) using standard @xmath88 distribution .
the bao peak parameter may be defined by @xmath89 where @xmath90 here , @xmath91 is the normalized hubble parameter ( i.e. , @xmath92 ) and @xmath93 is the typical redshift of the sdss data sample .
this quantity can be used even for more general models which do not present a large contribution of dark energy at early times @xcite .
now the @xmath11 function for the bao measurement can be written as in the following form @xmath94 where the value of the parameter @xmath95 for the flat model ( @xmath34 ) of the frw universe is obtained by @xmath96 using sdss data set @xcite from luminous red galaxies survey .
now the total joint data analysis ( stern+bao ) for the @xmath11 function defined by @xmath97 = 3 in = 2.5 in = 3 in = 2.5 in fig.3 fig.4 figs . 3 and 4 show the variation of @xmath3 with @xmath4 for @xmath76 and @xmath85 respectively for different confidence levels .
the 66% ( solid , blue ) , 90% ( dashed , red ) and 99% ( dashed , black ) contours are plotted for the @xmath23-@xmath9 ( stern)+ bao joint analysis .
now our best fit analysis with stern@xmath98bao observational data support the theoretical range of the parameters . in figures 3 and 4 , we plot the graphs for different confidence levels 66% ( solid , blue ) , 90% ( dashed , red ) and 99% ( dashed , black ) contours for @xmath76 and @xmath77 respectively and by fixing the other parameters .
the best fit values of @xmath78 , @xmath79 and the minimum values of @xmath80 are tabulated in table 3 .
@xmath99 * table 3 : * @xmath23-@xmath9 ( stern ) @xmath98 bao : the best fit values of @xmath78 , @xmath79 and the minimum values of @xmath80 for @xmath82 , @xmath83 , @xmath84 and for different values of @xmath8 .
another dynamical parameter that is used in recent cosmological tests is the cmb ( cosmic microwave background ) shift parameter which is the useful quantity to characterize the position of the cmb power spectrum first peak .
the cmb power spectrum first peak is the shift parameter which is given by @xcite @xmath100 where @xmath101 is the value of @xmath9 at the surface of last scattering .
wmap data gives @xmath102 at @xmath103 .
for cmb measurement @xmath11 function can be defined as @xmath104 and the total joint data analysis ( stern+bao+cmb ) for the @xmath11 function defined by @xmath105 = 3 in = 2.5 in = 3 in = 2.5 in fig.5 fig.6 figs .
5 and 6 show the variation of @xmath3 with @xmath4 for @xmath76 and @xmath85 respectively for different confidence levels . the 66% ( solid , blue ) , 90% ( dashed , red ) and 99% ( dashed , black ) contours are plotted for the @xmath23-@xmath9 ( stern)+ bao+ cmb joint analysis .
now our best fit analysis with stern @xmath98 bao @xmath98 cmb observational data support the theoretical range of the parameters . in figures 5 and 6
, we plot the graphs for different confidence levels 66% ( solid , blue ) , 90% ( dashed , red ) and 99% ( dashed , black ) contours for @xmath76 and @xmath77 respectively and by fixing the other parameters .
the best fit values of @xmath78 , @xmath79 and the minimum values of @xmath80 are tabulated in table 4 .
@xmath106 * table 4 : * @xmath23-@xmath9 ( stern ) @xmath98 bao @xmath98 cmb : the best fit values of @xmath78 , @xmath79 and the minimum values of @xmath80 for @xmath82 , @xmath83 , @xmath84 and for different values of @xmath8 .
= 3 in = 2.5 in fig.7 fig.7 shows the variation of distance modulus @xmath12 vs redshift @xmath9 for our model ( solid line ) and the union2 sample ( dotted points ) .
recent observation of high redshift survey supernovae type ia indicates that the universe undergoing an accelerating phase and gives an evidence of existence of dark energy . since 1995
, two teams of high - z supernova search and the supernova cosmology project have discovered several type ia supernovas at the high redshifts @xcite .
the observations directly measure the distance modulus of a supernovae and its redshift @xmath9 @xcite .
recent observational data , including sne ia which consists of the gold sample which has a 157 supernova @xcite and the another set is a combined data set of a 192 supernova @xcite . here
we consider 557 data points and belongs to the union2 sample @xcite which are considered in the next .
+ the luminosity distance @xmath107 and distance modulus @xmath12 for supernovas are calculated by @xmath108 and @xmath109 + 25\ ] ] the best fit of distance modulus as a function @xmath12 of redshift @xmath9 for our theoretical model and the supernova type ia union2 sample are drawn in figure 7 for our best fit values of @xmath78 , @xmath79 with the other chosen parameters . from the curves
, we see that the theoretical nvmcg model is in agreement with the union2 sample data .
we proposed here the frw universe filled with dark matter ( perfect fluid with negligible pressure ) along with new variable modified chaplygin gas ( nvmcg ) which is one of the candidates of dark energy .
we present the hubble parameter @xmath0 in terms of the observable parameters @xmath110 , @xmath111 with the redshift @xmath9 and the other parameters like @xmath78 , @xmath79 , @xmath112 , @xmath8 , @xmath6 and @xmath7 . we have chosen the observed values of @xmath113 and @xmath111 = 72 kms@xmath114 mpc@xmath114 . from stern data set
( 12 points ) , we have obtained the bounds of the arbitrary parameters @xmath78 and @xmath79 ( table 2 ) by minimizing the @xmath80 test and by fixing the other parameters @xmath115 and @xmath116 . in this way , we may found the bounds of any two parameters by fixing the remaining parameters .
next due to joint analysis of bao and cmb observations , we have also obtained the best fit values and the bounds of the parameters ( @xmath117 ) ( table 3 and 4 ) by fixing some other parameters @xmath118 and @xmath119 . the best - fit values and bounds of the parameters
are obtained by 66% , 90% and 99% confidence levels are shown in figures 1 - 6 for stern , stern+bao and stern+bao+cmb analysis .
the distance modulus @xmath12 against redshift @xmath9 has been drawn in figure 7 for our theoretical model of the nvmcg for the best fit values of the parameters and the observed sne ia union2 data sample . here
we show that our predicted theoretical nvmcg model permitted the observational data sets .
the observations do in fact severely constrain the nature of allowed composition of matter - energy by constraining the range of the values of the parameters for a physically viable nvmcg model . + +
* acknowledgement : * + the authors are thankful to iucaa , pune , india for warm hospitality where part of the work was carried out . also ud is thankful to csir , govt .
of india for providing research project grant ( no .
03(1206)/12/emr - ii ) .
+ 99 perlmutter , s. j. et al , 1998 , nature 391 , 51 .
perlmutter , s.j .
et al , 1999 , astrophys . j. 517 , 565 .
g. et al . , 1998 ,
j. 116 , 1009 .
tonry , j. l. et al . , 2003 , apj , 594 , 1 .
melchiorri a. et al.,2000,astrophysics j.lett.,536,l63 .
lange a.e .
et al.,2001,phys rev .
d , 63,042001 .
jaffe a.h .
et al.,2001,phys rev .
netterfield c.b .
et al.,2002,astrophysics , j.,571,604 .
halverson n.w .
et al.,2002,astrophysics .
briddle s. et al.,2003,science,299,1532 .
bennet c. et al.,2003,[astro - ph/0302207 ] .
hinshaw g. et al.,2003,[astro - ph/0302217 ] .
kogut a. et al.,2003,[astro - ph/0302213 ] .
spergel d.n .
et al.,2003,[astro - ph/0302209 ] .
a. kamenshchik et al , _ phys .
b _ * 511 * 265 ( 2001 ) .
gorini , v. , kamenshchik , a. and moschella , u. , 2003 , phys . rev .
d 67 , 063509 .
h. benaoum , hep - th/0205140 .
debnath , u. , banerjee , a. and chakraborty , s. , 2004 , class .
quantum grav .
21 , 5609 .
w. chakraborty and u. debnath , gravitation and cosmology , * 16 * 223 ( 2010 ) .
choudhury , t. r. and padmanabhan , t. , 2007 , astron .
429 , 807 .
lu , j. et al , 2008 , phys .
b 662 , 87 .
dao - jun , l. and xin - zhou , l. , 2005 , chin .
lett . , 22 , 1600 .
stern , d. et al , 2010 , jcap 1002 , 008 .
wu , p. and yu , h. , 2007 , phys .
b 644 , 16 .
thakur , p. , ghose , s. and paul , b. c. , 2009 , mon . not .
r. astron .
soc . 397 , 1935 .
paul , b. c. , ghose , s. and thakur , p. , arxiv:1101.1360v1 [ astro-ph.co ] .
paul , b. c. , thakur , p. and ghose , s. , arxiv:1004.4256v1 [ astro-ph.co ] .
ghose , s. , thakur , p. and paul , b. c. , arxiv:1105.3303v1 [ astro-ph.co ] .
eisenstein , d. j. et al , 2005 , astrophys .
j. 633 , 560 .
m. doran , s. stern and e. thommes , jcap 0704 , 015 ( 2007 ) .
o. elgaroy and t. multamaki , astron .
471 , 65e ( 2007 ) .
g. efstathiou and j. r. bond , mon . not .
304 , 75 ( 1999 ) .
a. g. riess et al .
, astrophys .
j. 659 , 98 ( 2007 ) ; astrophys .
j. 659 , 98 ( 2007 ) .
a. g. riess et al .
, astrophys .
j. 607 , 665 ( 2004 ) .
kowalaski et al , 2008 , astrophys .
j. 686 , 749 . | assuming the flat frw universe in einstein s gravity filled with new variable modified chaplygin gas ( nvmcg ) dark energy and dark matter having negligible pressure . in this research work
we analyze the viability on the basis of recent observation .
hubble parameter @xmath0 is expressed in terms of the observable parameters @xmath1 , @xmath2 and the model parameters @xmath3 , @xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 and the red shift parameter @xmath9 . here
we find a best fitted parameter range of @xmath3 , @xmath4 keeping @xmath10 and using stern data set ( 12 points ) by minimizing the @xmath11 test at 66% , 90% and 99% confidence levels .
next we do the joint analysis with bao and cmb observations . again evaluating the distance modulus @xmath12 vs redshift ( @xmath9 ) curve obtained in the model nvmcg with dark matter with the best fitted value of the parameters and comparing with that derived from the union2 compilation data . |
in the main test , we focus on the topological flat - band ( tfb ) model on honeycomb lattice and extract the modular matrix and related quasiparticle statistics through locating the minimal entangled states ( mess ) . in this supplemental material , we apply the similar route on checkboard lattice and we focus on searching the topological order of fractional quantum hall ( fqh ) state at filling factor @xmath89 @xcite . the hamiltonian for checkerboard lattice filled with hard - core bosons@xcite : @xmath154\nonumber\\ & \pm&t^{\prime}\sum_{\langle\langle\mathbf{r}\mathbf{r}^ { \prime}\rangle\rangle } \left[b^{\dagger}_{\mathbf{r}^{\prime}}b_{\mathbf{r}}+\textrm{h.c.}\right ] -t^{\prime\prime}\sum_{\langle\langle\langle\mathbf{r}\mathbf{r}^ { \prime}\rangle\rangle\rangle } \left[b^{\dagger}_{\mathbf{r}^{\prime}}b_{\mathbf{r}}+\textrm{h.c.}\right ] \nonumber\\ & + & v_1\sum_{\langle\mathbf{r}\mathbf{r}^{\prime}\rangle}n_{\mathbf{r}}n_{\mathbf{r}^{\prime}}\end{aligned}\ ] ] where @xmath155 creates a hard - core boson at site @xmath7 , @xmath156 is the boson number operator .
@xmath9 , @xmath10 and @xmath11 denote the nearest - neighbor ( nn ) , the next - nearest - neighbor ( nnn ) and the next - next - nearest - neighbor ( nnnn ) pairs of sites .
we adopt the parameters @xmath157 , @xmath158 , @xmath159 and @xmath160 , which leads to a tfb with the flatness ratio about @xmath161 . to stabilize the fqh phase at filling factor @xmath89 , we set nn interaction @xmath162 following the previous work@xcite .
we consider a @xmath163 checkboard lattice with five hard - core bosons . in the exact diagonalization study , there are four near degenerating eigenstates which are separated from higher eigenstates by a finite spectrum gap .
the four ground states @xmath87 ( @xmath164 ) lie in momentum sector @xmath165,@xmath166,@xmath167 and @xmath168 , respectively .
now we form the general superposition state from the four quasi - degenerating ground states , @xmath169 where @xmath170 are the real parameters and @xmath171 are the relative phase of the state respectively . for the bipartition along cut - i , we find that each groundstate @xmath87 is indeed the mes due to four ground states having different quantum number @xmath172 along the cut - i direction : @xmath173 for the partition along cut - ii , it is found that the mess appear when the four ground states are in equal magnitude superposition : @xmath174 . as shown in fig .
[ cb_hb ] , we show the entropy of wavefunction @xmath175 in @xmath176 space by setting @xmath174 . the color of dots represents the magnitude of entropy . for simplicity
, we just show the points with entropy smaller than @xmath177 .
it is clear that there exist four valleys in @xmath176 space .
the four valleys corresponding to four independent mess as : @xmath178 with state index @xmath164 and @xmath179 , @xmath180 , @xmath181 , @xmath182 , corresponding to minimum entropies @xmath183 , respectively .
the four mess are nearly orthogonal to each other : @xmath184 , @xmath185 , @xmath186 , @xmath187 , which forms orthogonal basis states for modular transformation . .
the arrow directions present the signs of the phases @xmath58 in the nn hopping terms .
the two ways to partition the system along the dashed lines are labeled as cut - i and cut - ii , respectively .
bottom : the entropy of wavefunction @xmath175 on @xmath163 checkboard lattice with @xmath188 interacting hard - core bosons by setting @xmath174 .
here we only show the entropy smaller than @xmath189 .
the calculation is for bipartition system along cut - ii direction .
, width=240 ] .
the arrow directions present the signs of the phases @xmath58 in the nn hopping terms .
the two ways to partition the system along the dashed lines are labeled as cut - i and cut - ii , respectively .
bottom : the entropy of wavefunction @xmath175 on @xmath163 checkboard lattice with @xmath188 interacting hard - core bosons by setting @xmath174 .
here we only show the entropy smaller than @xmath189 .
the calculation is for bipartition system along cut - ii direction . ,
width=288 ] as described in the main text , modular matrix can be obtained through the overlap between mess along two partition direction : @xmath111 @xcite . using eq.[mes : nu4:x ] and eq.[mes : nu4:y ] ,
we obtain , @xmath116 , which is nearly the same as @xmath117 statistics prediction @xcite up to @xmath190 correction : @xmath191 the modular matrix clearly demonstrates topological order of fqh states at @xmath89 . for example , from @xmath0 we determine : ( i)there are @xmath192 type quasiparticles in the system labeled by the charges @xmath193 , where @xmath194 ; ( ii)the quantum dimension of quasiparticles are all @xmath195 thus the total quantum dimension @xmath196 ; ( iii)the @xmath117 fusion rule : @xmath197 , where @xmath198 @xcite . | we perform an exact diagonalization study of the topological order in topological flat band models through calculating entanglement entropy and spectra of low energy states . we identify multiple independent minimal entangled states , which form a set of orthogonal basis states for the groundstate manifold .
we extract the modular transformation matrices @xmath0 ( @xmath1 ) which contains the information of mutual ( self ) statistics , quantum dimensions and fusion rule of quasiparticles .
moreover , we demonstrate that these matrices are robust and universal in the whole topological phase against different perturbations until the quantum phase transition takes place . _
introduction. _ the fractional quantum hall ( fqh ) state is the best - known many - body state with topological order discovered in 2d electron systems under strong magnetic field .
the most striking features of fqh state are the topological ground state degeneracy on torus and the emerging quasiparticles obeying fractional statistics @xcite .
recently , it has been demonstrated that fqh states can also be realized in various topological flat - band ( tfb ) models without landau levels @xcite .
in such an interacting system , an explicit demonstration of topological order and quasiparticle statistics is still highly desired , which has attracted lots of recent interests @xcite .
entanglement measurements such as topological entanglement entropy ( tee ) @xcite and entanglement spectrum @xcite have been identified as powerful tools for detecting topological properties of many - body quantum states .
insightfully , zhang _ et al .
_ proposed to extract modular matrix through the entanglement measurement @xcite , which encodes the complete information of the topological order including quasiparticles quantum dimension and statistics as first described by wen @xcite .
based on the model wavefunctions for toric code and chiral spin liquid states , they demonstrated that the transformation between the minimal entangled states ( mess ) along two interwinding partition directions gives rise to modular matrices .
the new route to extract modular matrix through mess improves the practical implementation for strongly interacting systems as such information is accessible through larger system density matrix renormalization group ( dmrg ) calculations demonstrated for bosonic fqh state @xmath2 in tfb @xcite and fermionic fqh states with magnetic field@xcite .
however , it remains difficult to access multiple low energy states in dmrg in a controlled way , when there are coupling between different topological sectors induced by interaction or when the groundstates have higher degeneracy , which will be the focuses of our exact diagonalization ( ed ) study . in this letter
, we present an ed calculation for the tfb model and map out the entanglement entropy profile for superposition states of the near degenerating groundstates .
we demonstrate that there are the same number of the mess as the ground state degeneracy for fqh phase on a torus , which form the orthogonal and complete basis states for modular transformation . through locating the mess along two interwinding partition directions
, we extract the modular matrices @xmath0 and @xmath1 containing generalized statistics of quasiparticles , which unambiguously demonstrate the fractional quasiparticle statistics in such systems for @xmath3 ( bosons ) , and @xmath4 ( fermions ) fqh states , respectively .
we also analyze the entanglement spectra and obtain tee from the difference of the maximum and minimum of entanglement entropies of these superposition states .
furthermore , we study the quantum phase transition from fqh phase to the topological trivial phase driven by the disorder scattering or attractive anisotropic interaction .
significantly , the extracted modular matrices remain to be universal containing the same quasi - particle fractional statistics information as theoretical ones for the model fqh states in the whole topological phase until the quantum phase transition takes place .
this is distinctly different from following the berry phase of the ground states , where only the sum of the total chern number remains invariant@xcite due to the lifting of the degeneracy by perturbations for any finite size systems .
we study the haldane model @xcite on the honeycomb ( hc ) lattice : @xmath5\nonumber\\ & -&t\sum_{\langle\mathbf{r}\mathbf{r}^ { \prime}\rangle } \left[c^{\dagger}_{\mathbf{r}^{\prime}}c_{\mathbf{r}}+\textrm{h.c.}\right ] -t^{\prime\prime}\sum_{\langle\langle\langle\mathbf{r}\mathbf{r}^ { \prime}\rangle\rangle\rangle } \left[c^{\dagger}_{\mathbf{r}^{\prime}}c_{\mathbf{r}}+\textrm{h.c.}\right]\nonumber \\ & + & v_1\sum_{\langle\mathbf{r}\mathbf{r}^{\prime}\rangle}n_{\mathbf{r}}n_{\mathbf{r}^{\prime } } + \sum_{\mathbf{r } } \epsilon_{\mathbf{r } } c^{\dagger}_{\mathbf{r } } c_{\mathbf{r } } \label{e.1}\end{aligned}\ ] ] where @xmath6 creates a hard - core boson ( or fermion ) at site @xmath7 , @xmath8 is the boson ( or fermion ) number operator .
@xmath9 , @xmath10 and @xmath11 denote the nearest - neighbor ( nn ) , the next - nearest - neighbor ( nnn ) and the next - next - nearest - neighbor ( nnnn ) pairs of sites , and @xmath12 is the nn interaction .
the last term models the anderson on - site disorder @xmath13 randomly distributed in @xmath14 $ ] . on the hc lattice ,
we select the parameters @xmath15 , @xmath16 , @xmath17 and the magnitude of the hopping phase @xmath18 , which lead to a topological flat - band with flatness ratio about @xmath19 @xcite .
we consider a finite system of @xmath20 unit cells ( total number of sites @xmath21 ) with periodic boundary conditions .
the filling factor is @xmath22 , where @xmath23 is the number of particles .
we denote the momentum vector @xmath24 with @xmath25 as integer quantum numbers .
the entanglement entropy is defined by partitioning the full system into two subsystems a and b. tracing out the subsystem @xmath26 , one can obtain the reduced density matrix of subsystem @xmath27 : @xmath28 , where @xmath29 is the many - body state of the full system .
the renyi @xmath30 entanglement entropy is defined as : @xmath31 . here
we focus on two noncontractible bipartitions on a torus , as shown in fig .
[ fig : hc:3d](a ) as cut - i and cut - ii , respectively .
_ multiple mess as superpositions of near degenerating groundstates. _ in tfb lattice model , it has been identified that there are @xmath32 near degenerating groundstates at filling factor @xmath33@xcite when the interacting system realizes a fqh phase .
let us first consider a @xmath34 hc lattice filled with hard - core bosons at half - filling @xcite .
we set the nn interaction to be zero since hard - core bosons are intrinsically interacting . from ed calculation
, we find the two groundstates @xmath35 and @xmath36 both in the same momentum sector @xmath37 .
this is the general case as long as the two system lengths @xmath38 and @xmath39 are factors of the particle number @xmath23 .
now we form the general superposition state as , @xmath40 and @xmath41 are the real parameter and the relative phase of the state respectively , while @xmath42 . for each state @xmath43 , we construct the reduced density matrix and obtain the corresponding entanglement entropy . in fig .
[ fig : hc:3d](b ) , we draw the @xmath44 in the surface and contour plots so that the peaks in entropy show up clearly representing the minimums of @xmath45 .
we identify two peak structures in @xmath46 parameter space corresponding to two independent mess : @xmath47 the minimal entropies at the two peaks are different with @xmath48 and @xmath49 respectively , indicating the finite size effect .
however , we find the relative phase difference between the two mess is @xmath50 and consequently the two mess are approximately orthogonal to each other : @xmath51 . due to the @xmath52 rotation symmetry in the @xmath34 system ,
the mess along cut - ii @xmath53 are related to @xmath54 as @xmath55 , where @xmath56 is the @xmath52 rotation operator .
hc lattice with lattice vectors @xmath57 . the arrow directions on red dotted lines present the signs of the phases @xmath58 in the nnn hopping terms .
the nnnn hoppings are represented by the blue dashed lines .
the two ways to bipartition the system along dashed lines are labeled as cut - i and cut - ii .
( b ) surface and contour plots of renyi @xmath30 entanglement entropy ( @xmath44 ) of wavefunction @xmath59 on @xmath34 hc lattice filled with @xmath60 hard - core bosons .
( c ) the entropy ( @xmath44 ) of wavefunction @xmath61 on @xmath62 hc lattice with @xmath60 interacting fermions .
, width=240 ] hc lattice with lattice vectors @xmath57 .
the arrow directions on red dotted lines present the signs of the phases @xmath58 in the nnn hopping terms .
the nnnn hoppings are represented by the blue dashed lines .
the two ways to bipartition the system along dashed lines are labeled as cut - i and cut - ii .
( b ) surface and contour plots of renyi @xmath30 entanglement entropy ( @xmath44 ) of wavefunction @xmath59 on @xmath34 hc lattice filled with @xmath60 hard - core bosons .
( c ) the entropy ( @xmath44 ) of wavefunction @xmath61 on @xmath62 hc lattice with @xmath60 interacting fermions .
, width=153 ] hc lattice with lattice vectors @xmath57 .
the arrow directions on red dotted lines present the signs of the phases @xmath58 in the nnn hopping terms .
the nnnn hoppings are represented by the blue dashed lines .
the two ways to bipartition the system along dashed lines are labeled as cut - i and cut - ii .
( b ) surface and contour plots of renyi @xmath30 entanglement entropy ( @xmath44 ) of wavefunction @xmath59 on @xmath34 hc lattice filled with @xmath60 hard - core bosons .
( c ) the entropy ( @xmath44 ) of wavefunction @xmath61 on @xmath62 hc lattice with @xmath60 interacting fermions .
, width=153 ] now we further examine the relation between the mess and the degeneracy of the ground state manifold by studying the tfb model filled with fermions at @xmath4 .
we consider a @xmath62 hc lattice with @xmath60 fermions with repulsive nn @xmath63 to stabilize the fqh phase @xcite . in the ed study
, we find three quasi - degenerating groundstates @xmath64 , ( with @xmath65 ) in momentum sectors @xmath66 , respectively .
we search for the superposition states in the space of the groundstate manifold with minimal entropy using the following general wavefunctions : @xmath67 are real parameters and @xmath68 , @xmath69 are relative phases for the state , while @xmath70 can be obtained using normalization condition . for the bipartition along cut - i , we observe two key points : 1 ) we can locate three global minimal entropy states in the given parameter space , which always occur when @xmath71 ; 2 ) the relative phases of two different mess @xmath72 and @xmath73 satisfy : @xmath74 , for @xmath75 . in fig . [
fig : hc:3d](c ) , we show the surface and contour plots of the entropy of the state @xmath76 as functions of @xmath68 and @xmath69 while other parameters are fixed at @xmath77 so that mess occur with varying the relative phases .
the three mess are determined as : @xmath78 with state index @xmath79 .
we find @xmath80 , @xmath81 , @xmath82 , corresponding to minimum entropies @xmath83 , respectively . very importantly ,
the three mess we found are nearly orthogonal to each other : @xmath84 , @xmath85 and @xmath86 , which is the necessary condition for these states to form the basis states for modular transformation .
the small overlap is a finite size effect as the mess become the true ground states only in the thermodynamic limit . since there is no rotation symmetry in @xmath62 lattice ,
we separately locate three mess in the parameter space for the partition along cut - ii . here , we find that each groundstate @xmath87 is indeed the mes @xmath88 in general , if the groundstates have different momentum along the entanglement cut direction , these states are eigenstates with definite number of quasiparticles , and thus any form of mixing will increase the entropy of the state .
we have also studied the tfb model on checkboard ( cb ) lattice @xcite and obtained similar results .
interestingly , we also identified a four fold degenerating mess at @xmath89 filling corresponding to a @xmath89 fqh@xcite . of the reduced density matrix of two mess ( square ) and two maximal entangled states ( diamond ) for ( a ) @xmath34 and ( b ) @xmath90 hc lattice with hard - core bosons at @xmath2 .
the number near the dots shows the degeneracy .
crossed blue ( purple ) dots stand for the combination of @xmath91 as described in the text.,scaledwidth=49.0% ]
.the comparison of the calculated tee @xmath92 ( for cut - i ) and the theoretical values @xmath93 for @xmath94 laughlin states .
hb and fm denote hard - core boson and fermion systems , respectively .
hc and cb represent honeycomb and checkboard lattices .
y(n) means the groundstates have the same ( different ) momentum . [ cols="<,^,^,^,^",options="header " , ] _ modular transformation matrix based on mess. _ the generalized quasiparticle statistics of a topological ordered state is captured by the modular matrix @xmath0 and @xmath1 as first proposed by wen @xcite .
@xmath0 ( @xmath1 ) determines the mutual ( self ) statistics of the different quasiparticles as well as the quantum dimension and fusion rules of quasiparticles @xcite . in general
, the relationship between the modular matrices and mess is @xmath95 , where @xmath96 are integers determined by specific modular transformation on a lattice @xcite .
specially , for a @xmath34 hc lattice filled with hard - core bosons , the @xmath52 rotation symmetry leads to the overlap @xmath97 @xcite .
thus by computing the overlap using states from eq .
[ mes : hc : boson ] , we obtain , @xmath98 which are nearly identical to the theoretical ones@xcite for the model fqh state : @xmath99 , @xmath100 . from @xmath101 , we determine the quantum dimension for two type of quasiparticles as @xmath102,@xmath103 and total quantum dimension @xmath104 ( close to @xmath105 ) .
@xmath106 show that one quasiparticle as a boson while @xmath107 indicates another quasiparticle acquires a @xmath108 phase encircling themselves . combined with the topological spin @xmath109 from @xmath110
, we identify that these quasiparticles are semions @xcite .
for @xmath62 hc lattice filled with interacting fermions , the overlap between eq .
[ mes : fermi : x ] and eq .
[ mes : fermi : y ] gives @xmath111 @xcite : @xmath112 the obtained result is close to the analytic prediction @xcite : @xmath113 , where @xmath114 . the extracted mutual statistics between quasiparticles reflects the @xmath115 statistics .
within the same route , we also obtain modular matrix for @xmath89 fqh states on cb@xcite : @xmath116 , which is nearly the same as the one representing @xmath117 statistics : @xmath118 .
_ topological entanglement entropy. _ for a topological ordered state , one can also identify a topological term in the entanglement entropy since @xmath119 , where @xmath120 is the length of the smooth boundary between two subsystems and the tee term is quantized as @xmath121 with @xmath122 as the total quantum dimension @xcite .
recently , it has been shown that tee of abelian fqh state can be extracted through @xcite , @xmath92 , where @xmath123 is the renyi @xmath30 entanglement entropy corresponding to maximal ( minimal ) entangled state . to check out this relation , the calculated @xmath92 for different systems
are shown in table [ table : gamma ] .
indeed , the obtained @xmath124 gives a good estimate of the quantized theoretical value @xmath125 for @xmath126laughlin state on torus @xcite .
for symmetric system of @xmath34 lattice , we obtained a bigger deviation between @xmath124 and @xmath127 , which may result from the strong coupling among the groundstates in the same momentum sector . to elucidate the physical difference between minimal and maximal entangled states , we further show entanglement spectra @xmath128 of these states in fig . [
fig : es : hc ] @xcite .
we find that the spectra of the maximal entangled states @xmath129 can be exactly recovered by reducing the density matrix eigenvalues by a factor @xmath32 ( @xmath130 for @xmath2 fqh state ) for two sets of spectra of mess and imposing them on top of each other : @xmath131 as shown as cross dots in fig .
[ fig : es : hc ] .
_ modular matrix and quantum phase transition. _ topological order is robust in the presence of any weak local perturbations , which can be used to characterize the topological phase .
here we first consider the disorder effect on bosonic state . as
shown in fig .
[ fig : hc : disorder](a - b ) , the energy spectrum remains two - fold quasi - degenerating protected by a spectrum gap until a disorder strength @xmath132 .
further calculation of particle entanglement spectrum ( pes ) reveals a gap at small @xmath133 and the number of states below this gap agrees with the number of quasihole excitations in a fqh state @xcite .
this pes gap disappears at @xmath134 signaling the quantum phase transition from the fqh phase to a topological trivial state . as shown in fig .
[ fig : hc : disorder](c - d ) , there are two distinguishable valleys i and ii in entropy for the states @xmath135 , and the corresponding mess are always approximately orthogonal to each other .
the modular matries obtained for an intermediate disorder strength @xmath136 are @xmath137 and @xmath138 , which remain to be very close to the exact results for bosonic @xmath2 laughlin state .
after the quantum phase transition at @xmath139 as shown in fig .
[ fig : hc : disorder](e ) , there are still two valleys of mess near @xmath140 and @xmath141 , however these two states start to have bigger overlap @xmath142 .
the corresponding modular matrix @xmath1 and @xmath0 are qualitatively different from exact results for @xmath2 fqh state as : @xmath143 and @xmath144 .
in particular , the quasi - particle statistics has changed with the @xmath1 matrix becomes unit matrix , which indicates we are in a topological trivial phase . at @xmath145
shown in fig .
[ fig : hc : disorder](e ) , there is only one valley left corresponding to one mes state in parameter space indicating the lost of any feature of topological order .
furthermore , we consider the effect of the anisotropic interaction for fermionic system by tunning the interaction @xmath146 on one nn bond , while keeping the other two at unit strength .
consistent with the geometrical theory of the fqhe @xcite , we find that the topological state and its modular matrix remain to be universal insensitive to the strength of the additional repulsive interactions with no quantum phase transition .
so we turn to the additional attractive interaction on one bond . from both energy spectrum and pes ,
we identify a quantum phase transition which appears between @xmath147 and @xmath148 as shown in fig .
[ fig : hc : disorder](g - h ) . as shown in fig .
[ fig : hc : disorder](i - l ) , in the fqh phase , there are three minimal entropy valleys in @xmath149 parameter space while we take @xmath150 , which are the optimized values for all these systems to minimize the entanglement entropy . in fqh phase ,
the calculated modular matrix is always nearly identical to the expected theoretic result for laughlin state .
taking @xmath151 as an example , from the overlap of the mess we extract @xmath152 .
after the phase transition occurs ( fig [ fig : hc : disorder](k - l ) ) at @xmath153 , we can only locate one minimal entropy valley in @xmath149 parameter space , which demonstrates the disappearance of the fqh phase .
_ summary and discussion. _ we study the structure of mess in the space of the groundstate manifold obtained from ed calculations . by calculating the overlap between different mess ,
we obtain modular matrices for different fqh systems .
the obtained @xmath0 and @xmath1 matrices faithfully represent the quasiparticle dimension and fractional statistics for systems with anisotropic interactions and random disorder scattering until a quantum phase transition takes place .
_ acknowledgements .
_ we thank shoushu gong for discussions .
this work is supported by the us doe office of basic energy sciences under grant no .
de - fg02 - 06er46305 ( dns ) and nsf under grants dmr-0906816 ( wz ) and the princeton mrsec grant dmr-0819860 ( fdmh ) .
dns also acknowledges the travel support by the princeton mrsec
. 99 r. b. laughlin , phys .
rev .
lett . * 50 * , 1395 ( 1983 ) .
x. g. wen , phys .
rev .
b * 40 * , 7387 ( 1989 ) .
f. d. m. haldane , phys .
rev .
lett . * 61 * , 2015 ( 1988 ) .
e. tang , j. w. mei and x. g. wen , phys .
rev .
lett .
* 106 * , 236802 ( 2011 ) .
t. neupert , l. santos , c. chamon and c. mudry , phys .
rev .
lett . * 106 * , 236804 ( 2011 ) .
k. sun , z. c. gu , h. katsura and s. das sarma , phys . rev .
lett . * 106 * , 236803 ( 2011 ) .
d. n. sheng , z. c. gu , k. sun and l. sheng , nature commun . * 2 * , 389 ( 2011 ) .
y .-
f .
wang , z .- c .
gu , c .- d . gong and d. n. sheng , phys .
rev .
lett . * 107 * , 146803 ( 2011 ) .
n. regnault and b. a. bernevig , phys .
rev .
x * 1 * , 021014 ( 2011 ) .
b. a. bernvig and n. regnault , phys .
rev .
b * 85 * , 075128 ( 2012 ) .
y. f. wang , h. yao , z. c. gu , c. d. gong , and d. n. sheng , phys . rev .
lett . * 108 * , 126805 ( 2012 ) .
y. l. wu , n. regnault and b. a. bernevig , phys .
rev .
lett . * 110 * , 106802 ( 2013 ) .
m. barkeshli and x .-
l .
qi , phys .
rev .
x * 2 * , 031013 ( 2012 ) .
t. scaffidi and g. moller , phys .
rev .
lett . * 109 * , 246805 ( 2012 ) .
d. xiao , w. zhu , y. ran , n. nagaosa , and s. okamoto , nat . commun . * 2 * , 596 ( 2011 ) .
g. murthy and r. shankar , phys .
rev .
b * 86 * , 195146 ( 2012 ) .
j. w. f. venderbos , s. kourtis , j. van den brink , and m. daghofer , phys .
rev .
lett . * 108 * , 126405 ( 2012 ) .
z. liu and e. j. bergholtz , phys .
rev .
b * 87 * , 035306 ( 2013 ) .
e. kapit and e. mueller , phys .
rev .
lett . * 105 * , 215303 ( 2010 ) .
s. yang , k. sun and s. das sarma , phys .
rev .
b * 85 * , 205124 ( 2012 ) .
s. a. parameswaran , r. roy , and s. l. sondhi , phys .
rev .
b * 85 * , 241308(r ) ( 2012 ) .
x. g. wen , int .
j. mod .
phys .
b * 4 * , 239 ( 1990 ) .
a. kitaev and j. preskill , phys .
rev .
lett . * 96 * , 110404 ( 2006 ) .
m. levin and x .- g .
wen , phys .
rev .
lett . * 96 * , 110405 ( 2006 ) .
h. li and f. d. m. haldane , phys .
rev .
lett .
* 101 * , 010504 ( 2008 ) .
a. m. lauchli , e. j. bergholtz , j. suorsa and m. haque , phys .
rev .
lett . * 104 * , 156404 ( 2010 ) .
h. c. jiang , z. h. wang and l. balents , nat .
phys .
* 8 * , 902 ( 2012 ) .
y. zhang , t. grover , a. turner , m. oshikawa and a. vishwanath , phys .
rev .
b * 85 * , 235151 ( 2012 ) .
l. cincio and g. vidal , phys .
rev .
lett . * 110 * , 067208 ( 2013 ) .
m. p. zaletel , r. s. k. mong , and f. pollmann , arxiv:1211.3733 .
h. h. tu , y. zhang , and x. l. qi , arxiv:1212.6951 .
d. n. sheng , x. wan , e. h. rezayi , k. yang , r. n. bhatt , and f. d. haldane , phys .
rev .
lett .
* 90 * , 256802 ( 2003 ) .
supplemental materials .
s. dong , e. fradkin , r. g. leigha and s. nowling , jhep 05 , 016 ( 2008 ) .
e. rowell , r. stong , z. h. wang , comm .
math .
phys . * 292 * , 343 ( 2009 ) .
p. fendley , m. p. a. fisher and c. nayak , j.stat.phys .
* 126 * , 1111(2007 ) .
t. grover , arxiv:1112.2215 .
a. m. lauchli , e. j. bergholtz and m. haque , new j. phys .
* 12 * , 075004 ( 2010 ) .
f. d. m. haldane , phys .
rev .
lett . * 107 * , 116801 ( 2011 ) .
b. yang , z. papic , e. h. rezayi , r. n. bhatt , and f. d. m. haldane , phys .
rev .
b * 85 * , 165318 ( 2012 ) . |
the importance of the study of possible contributors to the temperature fluctuations of the cosmic microwave background radiation is indisputable .
the large angular - scale anisotropy is dominated by the sachs
wolfe effect @xcite , where the first order perturbations in the metric tensor give rise to the appearance of temperature fluctuations in the cmbr .
in present paper we focus on the effects of rotational perturbations in a flat frw model filled with a single , incoherent , perfect fluid matter source , and we analyze its contributions to the fluctuations both in the absence and the presence of a cosmological constant .
this problem certainly has been studied by several authors before , eg .
@xcite , however none of these works payed too much attention on the possible analytic solutions in the presence of @xmath0 . throughout this paper
we follow a systematic treatment in calculating parallely with vanishing and nonvanishing cosmological constant and comparing the obtained results of the calculations . in section ii .
we take a general overlook on the @xmath1 solutions of the perturbed field equations given by sachs & wolfe @xcite , white @xcite and perjs et al .
@xcite , and the rotation tensor . in the presence of the cosmological constant we use the physical time coordinate @xmath2 in the calculations as the background solution has a closed analytic form only with this time variable while in the absence of @xmath0 we use the conformal time coordinate @xmath3 . in section iii .
we calculate the contributions of rotational perturbations to the temperature fluctuations due to the sachs
wolfe effect and analyze the obtained expression .
we also give a recipe in this section , how to calculate the anisotropy power spectra of the temperature fluctuations caused by rotational perturbations . in section iv .
, using the result of present observations , we give an upper estimation for the order of magnitude of the rotations and the corresponding angular velocity at the time of decoupling and today .
in the case of a perfect fluid matter source the einstein equations are @xmath4 with the four velocity @xmath5 of the fluid normalized by @xmath6 , @xmath7 is the pressure , @xmath8 is the energy density and @xmath9 is the einstein tensor .
the line element in a spatially flat frw spacetime with cartesian coordinate system is @xmath10 where @xmath3 is the conformal time coordinate with the definition @xmath11 , @xmath12 is the scale factor and @xmath13 is the minkowski metric with the following signature , @xmath14 here and throughout this paper roman indices run from 0 to 3 , greek indices from 1 to 3 and we employ units such that @xmath15 . the well - known unperturbed solution of the einstein equations in the case of incoherent matter without cosmological constant is @xmath16 where @xmath17 is the present value of the hubble parameter @xmath18 , the prime denotes the derivative with respect to the conformal time and we chose the @xmath19 normalization condition . the also well - known background solution of the field equations for incoherent matter in the case of @xmath20 is @xmath21 where @xmath22 is restricted by the normalization condition of @xmath23 .
hereafter in this section we follow the perturbation method of sachs & wolfe @xcite and white @xcite and we write the perturbed metric @xmath24 as @xmath25 where @xmath26 is the metric perturbation .
we use the minkowski metric @xmath13 and its inverse to raise and lower indices of @xmath26 and of other small quantities . in the perturbed spacetime
we choose comoving coordinates , where the four velocity of the fluid and the metric perturbation @xmath27 are @xmath28 the general @xmath1 solution of the perturbed field equations for incoherent matter without cosmological constant is given by sachs & wolfe @xcite and white @xcite , @xmath29 where and throughout this paper comma denotes partial derivative and @xmath30 , @xmath31 are two scalar functions corresponding to the potentials for density perturbations .
the @xmath32 vector function satisfying the @xmath33 transversality condition is related to the perturbed rotation ( vorticity ) tensor @xmath34 , see ( [ rot ] ) .
the @xmath35 transverse , trace - free tensor function represents gravitational waves as the solution of the following wave equation : @xmath36 where @xmath37 is the standard laplacian defined by @xmath38 , for an arbitrary smooth function @xmath39 .
the general @xmath1 solution of the perturbed field equations for incoherent matter in the case of @xmath40 is given by perjs et al .
@xcite , @xmath41 where @xmath42 and @xmath43 time - dependent amplitudes are the following elliptic integrals , @xmath44 which can be brought to the legendre normal form given in the appendix of @xcite .
the @xmath45 represents gravitational waves analogous to the term containing @xmath35 in ( [ nolsol ] ) .
we note that in eqs.([hab ] ) , ( [ pertl ] ) and ( [ j ] ) the integration functions @xmath46 and @xmath43 are defined with an opposite sign compared to the original results of perjs et al . in @xcite .
this sign convention helps us in the following to show the clear correspondence of the forthcoming results derived parallely from the sachs & wolfe and the @xmath20 solutions . in the chosen comoving coordinate system ,
the observers move together with the perturbations of the fluid in the perturbed spacetime . using the expression for the four velocity @xmath47 given in ( [ conv ] ) we can calculate the rotation tensor @xmath34 defined by ehlers @xcite @xmath48 where @xmath49 and the semicolon denotes the covariant derivative with respect to the perturbed metric @xmath24 .
the four velocity @xmath50 is connected to the @xmath51 components of the metric perturbation from which with the use of the solution of the field equations in the case of vanishing cosmological constant we have the following expression for @xmath34 @xmath52 repeating the calculations in the case of @xmath20 we have @xmath53 the relations above demonstrate that the perturbed vorticity tensor is completely determined by the @xmath54 integration function .
in present section we calculate the contribution of rotational perturbations to the fluctuations of the cosmic microwave background radiation . as a consequence of the sachs wolfe effect @xcite , the perturbations produce contributions to the temperature fluctuations determined by the following relation @xmath55 with @xmath56 here the vector @xmath57 represents the spatial direction of the light signal as seen by an observer moving together with the fluid and the @xmath58 subscript denotes the instant of time of the emission . applying the following identity for an arbitrary smooth function @xmath59 @xmath60 and calculating only with rotational perturbations ( terms containing @xmath46 in the perturbed solutions ) , ( [ dett1 ] ) can be written in the following general form , @xmath61_0^{\eta_r-\eta_e}-\int_0^{\eta_r-\eta_e}\xi(\eta ) c_{\alpha}e^{\alpha } dw\ , \end{aligned}\ ] ] were the @xmath62 , @xmath63 and @xmath64 time dependent amplitudes are the following in the cases of vanishing and nonvanishing cosmological constant respectively : @xmath65 in fig.1 we have plotted the @xmath62 , @xmath63 and @xmath64 time dependent amplitudes on a logarithmic scale .
the shape of the curves show that the presence of the cosmological constant results only about a @xmath66 factor difference from the single fluid model at each functions .
this means that the fluctuations represented by rotational perturbations in the microwave background are @xmath67 times smaller in the presence of @xmath0 .
it is also remarkable that the magnitude of the amplitudes are decreasing exponentially since the time of decoupling , and the present day contributions are negligible comparing to early epochs . except the integrated term in ( [ gf ] )
whose contribution is growing in time the contributions of the first two terms are the difference of their values at the time of decoupling and today .
therefore the dominant parts come from the time of decoupling and as the present values are many orders of magnitude smaller these terms produce the only relevant contributions to the temperature fluctuations that could be measured today . the time dependence of the @xmath68 , @xmath69 and @xmath70 functions in the presence and the absence of @xmath0 .
the subscript @xmath0 denotes the corresponding functions in the model with cosmological constant .
the vertical lines belong to the following events : @xmath71 and @xmath72 denote the time of decoupling in the models with and without cosmological constant , while the @xmath73 and @xmath74 denote the present age of the universe in the two models respectively . ] in the remainder of this section we present the recipe of the calculations how to obtain the anisotropy spectra of the fluctuations generated by rotational perturbations . as a first step , following bardeen s covariant formalism , we replace the @xmath75 perturbation function with a single mode of its decomposition into the vector harmonics @xmath76 of the helmholtz equation @xcite @xmath77 where the subscript @xmath78 denotes the wavenumber dependence of the coefficients @xmath79 and @xmath80 the spatial unit vectors @xmath81 and @xmath82 are spanning the plane perpendicular to @xmath83 . applying this expansion ,
the integrated term in ( [ gf ] ) can be written in the following form in the case of vanishing cosmological constant @xmath84 using the @xmath85 replacement one can evaluate this isw integral in a closed analytic form as follows , @xmath86\ .\end{aligned}\ ] ] repeating the calculations in the case of @xmath40 , the integrated term in ( [ gf ] ) takes the following shape @xmath87 where @xmath88 and @xmath89 the isw(@xmath0 ) integral in ( [ iswl ] ) can not be evaluated analytically thus numerical integration needs to be used . to be able to obtain the anisotropy power spectra generated by rotational perturbations
, one has to evaluate the following well - known integral @xmath90 where @xmath91 is the anisotropy momenta and @xmath92 is identical with @xmath93 in ( [ gf ] ) in the decomposition by vector harmonics of the helmholtz equation presented above . as the last step , one has to take the square of the absolute value of @xmath94 given in ( [ gf ] ) . to be able to evaluate the integral expression of @xmath91
, one needs to have information about the spectrum of @xmath95 which might be originated from predictions of models describing the universe before the decoupling era . as an example
this has been modelled by grishchuk in @xcite , where he derives the angular variation of the microwave background radiation from rotational perturbations of quantum - mechanical origin .
as an application we estimate the magnitude of the angular velocity generated by rotational perturbations using the results of present observations of the cmbr fluctuations .
it is well - known from the analysis of recent measurements @xcite that the dipole component of the temperature fluctuations has the largest amplitude with the order of magnitude @xmath96 .
this term certainly contains many contributions , however one of these originates from the effect of rotational perturbations . as we have shown in ( [ gf ] ) this contribution can be generally separated into the following three components : one proportional to @xmath46 , one proportional to @xmath54 and an integrated term proportional also to @xmath46 . since the rotation tensor is completely determined by the @xmath54 function , we must estimate the order of magnitude of the @xmath54 perturbation to be able to obtain the magnitude of the corresponding angular velocity vector @xmath97 . applying the results of present observations of the dipole component , we can make an estimation on the upper bound of the rotations with the following procedure .
the corresponding angular velocity deduced from the vorticity tensor is @xmath98 where @xmath99 and @xmath100 is the 3-dimensional totally antisymmetric levi - civita symbol .
having done the calculation for the square root of @xmath101 , we obtain the expression @xmath102^{1/2},\end{aligned}\ ] ] and the @xmath103 time dependent amplitude is the following in the cases of vanishing and nonvanishing cosmological constant respectively , @xmath104 as an upper limit , in accordance with the measurements , we can say that none of the terms in ( [ gf ] ) can be larger than the order of magnitude @xmath105 , so this must also be valid for the @xmath54 term . in the previous section we have shown that the relevant contribution of this term comes from its value at the time of decoupling , where the magnitude of its time dependent amplitude @xmath106 is known .
this makes the restriction to @xmath54 that its order of magnitude can not be larger than @xmath107 .
if we assume that the first derivative @xmath108 has the same order of magnitude , we can make the estimation for @xmath109 in ( [ kappa ] ) that is @xmath110 . applying this assumption
we obtain the following order of magnitudes for the values of the angular velocity at the time of decoupling and today in the presence and the absence of @xmath0 : @xmath111 it shows that the presence of the cosmological constant does nt change the order of magnitude of the angular velocity , however in a more accurate calculation there might be a multiplying factor @xmath112 between the two models similar to the results obtained for @xmath94 in the previous section .
the present value of @xmath113 can be associated with the rotation of the universe , and the obtained order of magnitudes are in good agreement with the results of @xcite , where the estimated value for the ratio @xmath114 in a flat model .
we have considered the effects of rotational perturbations of frw models on the cosmic microwave background radiation and its connection to the amplitude of the rotation of the universe .
we presented the general form of the relative temperature fluctuations induced by rotations and obtained analytic expressions for all the time dependent coefficients arising both in the presence and the absence of a cosmological constant .
it is found that the @xmath62 and @xmath63 transfer functions of the rotational sachs
wolfe effect , and the @xmath64 weighing function of the rotational integrated sachs
wolfe effect are decaying exponentially , and there is a @xmath66 factor between their values in the models with zero and nonzero cosmological constant .
we presented the general method of calculating the anisotropy power spectra of the rotational isw term , and using the results of recent observations for the dipole component of the microwave background , we made an upper estimation for the magnitude of the rotation of the universe at the time of decoupling and today .
it is a remarkable result that the value of the corresponding angular velocity at the time of decoupling is close to the order of magnitude of the angular velocity of a typical galaxy cluster today @xcite .
this interesting coincidence might deserve attention in future investigations .
we thank rpd lukcs and lszl b. szabados for useful conversations and for their help in preparing the manuscript .
we are also thankful to lajos balzs for his valuable help to find our way in the astronomical literature .
this work was supported by jsps p06816 and otka no .
f049429 grants .
we present the list of physical parameters that have been used throughout the calculations obtaining numerical values .
@xmath115 [ sec : texbooks ] r. k. sachs , a. m. wolfe , , * 147 * , 73 ( 1967 ) .
a. m. anile , s. motta , astron .
astrophys , * 32 * , 137 ( 1974 ) .
j. d. barrow , r. juszkiewicz and d. h. sonoda , mon . not .
r. astron .
* 213 * , 917 ( 1985 ) .
l. p. grishchuk , , * 48 * , 5581 ( 1993 ) .
p. c. white , j. math .
* 14 * , 831 ( 1973 ) .
z. perjs , m. vasth , v. czinner and d. eriksson , astron .
astrophys , * 431 * , 415 ( 2005 ) .
j. ehlers , akad .
mainz , * 11 * , 812 ( 1961 ) . j. m. bardeen , , * 22 * , 1882 ( 1980 ) . c. j. copi , et al .
, mon . not .
r. astron .
soc * 367 * , 79 ( 2006 ) g. de vaucouleurs , publ .
pacific * 83 * , 113 ( 1971 ) m. kalinkov , et al . , mon . not .
r. astron .
soc * 359 * , 1491 ( 2005 ) | we consider general - relativistic rotational perturbations in homogeneous and isotropic friedmann robertson walker ( frw ) cosmologies .
taking linear perturbations of frw models , the general solution of the field equations contains tensorial , vectorial and scalar functions .
the vectorial terms are in connection with rotations in the given model and due to the sachs wolfe effect they produce contributions to the temperature fluctuations of the cosmic microwave background radiation ( cmbr ) . in present paper we obtain the analytic time dependence of these contributions in a spatially flat , frw model with pressureless ideal fluid , in the presence and the absence of a cosmological constant @xmath0 .
we find that the solution can be separated into an integrable and a non - integrable part as is the situation in the case of scalar perturbations .
analyzing the solutions and using the results of present observations we estimate the order of magnitude of the angular velocity corresponding to the rotation tensor at the time of decoupling and today . |
hot accretion flow , such as advection - dominated accretion flows ( adafs ; ichimura 1977 ; rees et al .
1982 ; narayan & yi 1994 ; 1995 ; abramowicz et al . 1995 ; see narayan , mahadevan & quataert 1998 ; kato , fukue & mineshige 1998 ; and yuan & narayan 2013 for reviews ) , is of great interest because of its widespread applications in low - luminosity agns , including the spuermassive black hole in our galactic center , and the quiescent and hard states of black hole x - ray binaries ( see reviews by narayan 2005 ; yuan 2007 ; narayan & mcclintock 2008 ; ho 2008 ; and yuan 2011 ) . in the early analytical works , the mass accretion rate is assumed to be independent of radius , @xmath13 . in this case , the radial profile of density satisfies @xmath14 ( e.g. , narayan & yi 1994 ) . numerous hydrodynamical ( hd ) and magnehydrodynamical ( mhd ) numerical simulations have been done , with most of them focusing on the dynamics of the accretion flow ( e.g. , igumenshchev & abramowicz 1999 , 2000 ; stone , pringle & begelman 1999 , hereafter spb99 ; stone & pringle 2001 ; hawley , balbus & stone 2001 ; machida , matsumoto & mineshige 2001 ; mckinney & gammie 2002 ; hawley & balbus 2002 ; igumenshchev , narayan & abramowicz 2003 ; pen , matzener & wong 2003 ; de villiers , hawley & krolik 2003 ; proga & begelman 2003a , 2003b ; pang et al . 2011 ; mckinney , tchekhovskoy & blandford 2012 ; narayan et al .
the effect of strong radiation was studied by yuan & bu ( 2010 ) , and most recently by li , ostriker & sunyaev ( 2012 ) .
one of the most surprising which is perhaps also the most important findings of these simulations is that the mass accretion rate ( or precisely the inflow rate ; refer to eq .
[ 7 ] in the present paper for its definition ) , is found to be not a constant ; but rather , it decreases with decreasing radius . denoting the mass accretion rate as @xmath15 ,
numerical simulations have found that @xmath16 ( see 4.1 for a review ) .
consequently , the density profile flattens compared to the previous @xmath17 : we now have @xmath18 with @xmath19 .
such results have obtained strong observational supports in the case of the supermassive black hole in our galactic center , sgr a * ( yuan , quataert & narayan 2003 ; refer to 4.2.1 for details . ) , the low - luminosity agn ngc 3115 ( wong et al . 2011 ; refer to 4.2.2 for details ) , and black holes in elliptical galaxies ( di matteo et al .
2000 ; mushotzky et al .
2000 ) .
in addition to the obvious theoretical interest , the radial profiles of accretion rate and density have also important observational applications .
this is because they will determine the emitted spectrum and other radiative features of an accretion flow ( e.g. , quataert & narayan 1999 ; yuan , quataert & narayan 2003 ) .
for example , in the case of sgr a * , the mass accretion rate at the bondi radius can be determined directly by observations .
then depending on the radial profile of the accretion rate ( or more exactly the density ) , the radiation of the accretion flow is completely different . the radial profile of accretion rate also determines the evolution of black hole mass and spin .
in many current numerical simulations , due to resolution difficulty , we can at most resolve the bondi radius and determine the bondi accretion rate there .
then the evolution of mass and spin of black holes will be determined by the fraction of the bondi accretion rate that finally falls onto the black hole , which is determined by the radial profile of accretion rate .
it is thus important to carefully investigate the radial profiles of accretion rate and density .
one problem of almost all previous simulations is that the radial dynamical range is rather small , usually at most two orders of magnitude .
this is technically because it would be so time - consuming that it is almost impossible to simulate an accretion flow , if the dynamical range is too large .
in addition , as is well known the simulation results are usually not reliable close to the boundary due to the boundary condition effects .
these cast some shade on the previous simulation results , especially the exact quantitative radial profiles of the physical quantities .
the situation is even worse for mhd simulations compared to hd ones . because the alfvn speed is very large in the region of low density and strong magnetic field , mhd simulation is much more expensive than hd simulations .
the radial dynamical range is thus more limited , and it is harder to evolve the simulation for long time which further constraints the range over which a steady state is reached .
a large dynamical range is also useful to investigate the following problem .
previous works ( spb99 ; igumenshchev & abramowicz 1999 ; yuan & bu 2010 ) have found that the bernoulli parameter of most outflow in their simulations is indeed negative
. one may expect that outflow may not be able to escape to infinity , but may rejoin the accretion flow at a certain distance .
the fact that all previous simulations never found the accumulation of matter in the accretion flow could be because of two possible reasons .
one is that the simulation time is not long enough for accumulation to occur .
however , the radial velocity of outflow is roughly in the order of the local keplerian velocity ( paper ii ) .
we can therefore expect that if the outflow would finally rejoin the accretion flow , the timescale required for accumulation would be shorter than the accretion timescale by a factor of @xmath20 .
thus this possibility is unlikely .
another possible reason is that the radial dynamical range is too small .
the accumulation may occur at a radius larger than the outer boundary of all current simulations .
to examine this possibility , a larger dynamical range is required . in the present paper
we simulate two - dimensional hd accretion flows .
especially , a `` two - zone '' approach will be adopted which helps us to overcome the technical problem and achieve a large dynamic range spanning four orders of magnitude .
it is widely believed that in reality the magnetic stress associated with the mhd turbulence driven by the magnetorotational instability ( mri ) transfers angular momentum ( balbus & hawley 1991 ; 1998 ) .
therefore , we should in principle use mhd simulation . in the present paper
we do nt include magnetic field but instead include an anomalous shear stress to mimic the magnetic stress .
however , as we will describe in paper ii , in hd and mhd accretion flows , the mechanisms of producing the accretion rate profile are different .
one may ask whether the radial profile of accretion rate obtained in the present hd simulation is the same with that obtained in more realistic mhd simulations . in this regard
, the recent work by begelman ( 2012 ) gives a positive answer .
the comparison between our hd simulation with mhd simulations also confirms this point ( refer to 4.1 for details ) .
the structure of the present paper is as follows . in 2 , we introduce the details of our `` two - zone '' simulation approach .
the results of simulations are presented in 3 .
we find that the profiles of the mass accretion rate and density still follows a perfect power - law form and the power - law index almost remains unchanged compared to previous results . in
4 we compare our results with previous hd and mhd simulation works ( 4.1 ) and observations ( 4.2 ) .
we find broadly good agreement among them .
the last section ( 5 ) devotes to a summary .
the equations describing the hydrodynamics of accretion flows are taken from spb99 : @xmath21 @xmath22 @xmath23 we use the spherical coordinates @xmath24 to solve these equations . in the above equations , @xmath25 , @xmath12 , @xmath26 , and @xmath27 are the mass density , pressure , velocity , and internal energy density , respectively . here
we adopt an adiabatic equation of state @xmath29 with @xmath30 .
@xmath31 is the gravitational potential .
we adopt the paczysky & wiita ( 1980 ) potential @xmath32 , where @xmath33 is the center black hole mass , @xmath34 is the gravitational constant and @xmath35 .
the self gravity of the accretion flow is neglected . in a real accretion flow , the maxwell stress associated with the mhd turbulence driven by the magnetorotational instability transfers the angular momentum . in the present work ,
we do not include magnetic field ; instead we follow spb99 and add the final terms in equations ( 2 ) and ( 3 ) to represent the anomalous shear stress and the corresponding heating . here
@xmath36 is anomalous stress tensor . to approximate the effect of magnetic stress , again following spb99 , we assume that only azimuthal components of the stress are non - zero because mri is driven only by the shear associated with the orbital dynamics : @xmath37 @xmath38 this approximation is supported by three - dimensional mhd simulations ( e.g. , hawley , gammie & balbus 1996 ; stone et al .
we emphasize this approximation because the inclusion of other components was found to produce quite different results ( igumenshchev & abramowicz 1999 ) . the viscosity coefficient @xmath39 determines the magnitude of the stress and @xmath40 is the kinematic viscosity coefficient .
we adopt the form @xmath41 because it corresponds to the usual `` @xmath20 '' description ( spb99 ) . in the present paper
we investigate two cases , i.e. , @xmath20=0.001 ( model a ) and @xmath20=0.01 ( model b ) .
we choose units such that @xmath42 and @xmath43 .
we aim at simulations with a large dynamical range of four orders of magnitude in radius .
directly simulating such a large range is technically almost impossible . for this purpose
, we divide the whole simulation domain into two zones .
the inner zone is from 1.35 @xmath0 to @xmath44 while the outer one from 100 @xmath0 to 40000 @xmath0 .
we simulate these two zones separately , following the four steps described below . for each zone
, we use the zeus-2d code ( stone & norman 1992 ) in spherical geometry to solve the equations .
the two modifications to the code are to include the shear stress terms and the implementation of the paczysky & wiita ( 1980 ) potential .
we adopt the same non - uniform grid in both the radial and angular directions as in spb99 .
the resolution is @xmath45 and @xmath46 . _ step one_. we first simulate the outer zone . following spb99 , the initial state of our simulation is an equilibrium torus with a constant specific angular momentum .
the readers are referred to spb99 for the description of the torus .
we set the radius of the maximum density at @xmath47 , the maximum density of the torus @xmath48 , the density of the medium @xmath49 , and pressure @xmath50 .
the standard outflow boundary conditions ( projection of all dynamical variables ) at both the inner and outer boundaries are adopted .
_ step two .
_ we then simulate the inner zone .
we inject the gas at the outer boundary .
the values of density , specific internal energy , and velocity of the injected flow are taken from the steady simulation results of step one at @xmath51 .
we do not use @xmath52 because we hope to avoid the effect of the inner boundary .
we adopt the outflow boundary condition at the inner boundary .
_ step three .
_ obviously , simply connecting the simulation results of the last two steps is generally not self - consistent .
this is because , for example , the inner zone will obviously produce some outflow at the outer boundary and these outflow will be injected into the outer zone .
this effect has not been taken into account in step one .
we therefore simulate the outer zone again , but this time we use the results of step two at @xmath53 as the inner boundary condition . again
, here we do not choose @xmath44 because we hope to avoid the effect of outer boundary .
in this way , we can capture all the outflow produced from the inner region and preserve all their properties such as their velocity and internal energy , and thus their ( negative ) bernoulli parameter .
thus , we should be able to observe whether these outflow will rejoin the accretion flow and accumulate somewhere in the outer zone .
after the outer zone reaches the steady state , we simulate the inner zone again , following the method described in step two .
_ step four .
_ we plot the radial profiles of various physical quantities along different @xmath54 of the inner and outer zones .
we by - eye look at the curves in each zone to judge the convergence , i.e. , whether the results in each zone obtained in two adjacent steps are consistent with each other . if not , we do iteration to improve
. usually we can get satisfactory convergence after up to three iterations .
however , we note that this iteration approach does not guarantee a complete consistency between the two zones . in principle
, we should transfer all information of the fluid at each time step from one zone to another .
we assume steady solution is reached and only transfer part of the information at one snapshot .
however , our approach does transfer some information between the two zones , and these information is perhaps the most important for the dynamics of accretion flow . using this `` two - zone '' approach
, we can easily simulate the accretion flow with a dynamic range spanning four orders of magnitude in radius .
compared to the usual `` one - zone '' approach , the calculation time costed is about @xmath55 several hundreds times shorter for our problem ( here @xmath56 is the number of iterations ) .
this approach could potentially be used in other simulation problems which require a large dynamical range .
but we should emphasize that the underlying assumption of this approach is that steady solutions exist .
this is satisfied for our problem .
figure 1 shows the time - averaged radial distribution of physical quantities near the equator of the whole region of the accretion flow ( both the inner and outer zones ) .
the results are averaged over time and angle between @xmath57 and @xmath58 .
the solid lines are the final results of _ step four_. for comparison aim we also show by the dot - dashed lines the results of _ step one _ and _ step two_. we can see that the profiles slightly become flatter after convergence is achieved . the density , gas pressure , rotation velocity and radial velocity can be described by a power law scaling with radius , r^-0.65 , pr^-1.7 , v _
r^-0.5 , v_r r^-0.55 .
[ radialscaling]these scalings are measured away from the innermost region , i.e. , @xmath59 .
we avoid the region within @xmath8 because of two reasons .
firstly , as emphasized by narayan et al .
( 2012 ) and also confirmed by the radial velocity plot in figure 1 , close to the black hole , the radial velocity of the flow increases inward much more rapidly because of the strong gravity , thus it deviates from its scaling extrapolated from the region of @xmath59 .
secondly , as we will see from figure 2 below , within @xmath7 , the inflow rate is a constant and there is little outflow .
therefore , in this sense this region is also `` special '' compared to the region outside @xmath7 .
we note that these two `` inner boundary effects '' do not exit if a newtonian potential is adopted .
these results are roughly consistent with spb99 , where they found @xmath60 , @xmath61 and @xmath62 .
we have done some test calculations and found that the small discrepancy is because we adopt the paczysky & wiita potential while a newtonian potential is adopted in spb99 .
the scaling of the radial and rotational velocities are also consistent with the self - similar solution of adaf ( narayan & yi 1994 ) , where we have @xmath63 .
the deviations of @xmath25 and @xmath64 from the self - similar solution are significant .
this is because of the inward decrease of mass accretion rate , as we will describe below .
we note that the scaling we obtain should be more reliable than previous works because of our extremely large radial dynamical range . in figure 2 , we plot the time - averaged and angle - integrated mass accretion rate of the whole region . following spb99 , the mass inflow and outflow rates , @xmath65 and @xmath66 , are defined as follows , @xmath67 @xmath68 the net mass accretion rate is defined as , @xmath69 the rates of inflow and outflow , and the net rate are denoted by the solid , dashed , and dotted lines , respectively .
note that the results are obtained by time - average the integral rather than integrating the time - averages . also shown in the figure by the dot - dashed line
is the inflow rate obtained in _
step one _ and _ step two_. we see that the final curves of inflow rate after _ step four _ becomes steeper compared to the results of _ step one _ and _ step two_. the net rate close to the outer boundary of the outer zone is not constant , indicating that it requires longer time to get the fully steady solution there .
this is because the viscosity coefficient is very small so the accretion timescale is very long .
the radial profile of the inflow rate from @xmath70 to @xmath71 can be described by _
in(r ) = _ in(r_out)()^0.48.[inflowprofile]or more precisely by , _ in(r)= _
in(r_out)()^0.65[inflowprofile2]from @xmath7 to @xmath71 , while it is almost constant within @xmath8 .
this is different from spb99 . in that work
, they found the inflow rate keeps decreasing inward until the inner boundary , and the slope is steeper than ours , which is @xmath72 ( `` run k '' in spb99 ) .
again , the reason is because we adopt a different gravitational potential of the black hole .
actually , we can see from the figure that , beyond @xmath52 where the two types of potential is basically identical , the power - law index is @xmath73 , in good consistency with spb99 .
the deeper potential adopted in our work suppresses convection to some degree and makes the outflow weaker . in paper ii
, we will analyze the origin of the inward decrease of the accretion rate .
we find that it is because of the mass loss in the outflow , and the the production of outflow is because of the gradient of gas pressure , i.e , the buoyant force .
when the gravitational force is stronger , the gas pressure force plays a relatively minor role .
in other words , in the innermost region , due to the strong gravity the accretion timescale is shorter than the timescale required for the formation of outflow thus few outflow is produced .
if the flat density profile of @xmath74 is because of the inward decrease of mass accretion rate , we should expect that the power - law index @xmath12 ( @xmath3 ) should be related to @xmath11 ( @xmath75 ) by @xmath76 . from equations ( [ radialscaling ] ) and ( [ inflowprofile2 ] )
, we have @xmath77 .
the small deviation is mainly because the density profile depends not only on the inflow profile , but on the sum of the inflow and outflow rates , i.e. , @xmath78 .
the radial profile of outflow rate is steeper than that of the inflow rate , _ out(r)r^0.87throughout the radius or _ out(r)r^0.8for the region beyond @xmath79 . at large radii , @xmath80 ,
the profiles of inflow and outflow rates are almost identical thus the relation @xmath76 is better satisfied , where @xmath81 and @xmath82 .
the most notable result from figures 1 & 2 is that the profiles of physical quantities , such as density and mass inflow rate , are well described by a single power - law form , extending from @xmath71 to @xmath83 .
this confirms previous simulation results such as spb99 , although in those simulations the dynamical range of the whole simulation domain usually spans only two orders of magnitude , and the steady solution is only reached within one order of magnitude because of the effect of outer boundary condition .
this result indicates that although the bernoulli parameter of most of the mass outflow is negative ( spb99 and yuan & bu 2010 ) , the outflow does not rejoin the accretion flow and accumulate somewhere within @xmath84 .
physically , this is likely because the flow is viscous thus there is an associated energy flux .
we simulate model b to examine the effect of varying the amplitude of the shear stress . the viscosity coefficient in model
b is 10 times larger than model a. igumenshchev & abramowicz ( 1999 ) and spb99 have already investigated this issue .
they found that when the stress becomes stronger , the convective instability becomes weaker , in the sense that the radial profile of inflow rate becomes flatter .
but they did not present the qualitative results of the profiles of density and inflow rate when the stress is increased . in the extreme case , igumenshchev & abramowicz ( 1999 ) found that when @xmath86 , the flow becomes almost laminar , and a bipolar outflow structure very close to the rotation axis , @xmath87 , is produced .
in the present work , for the aim of comparing with observations ( see 4 below ) , we try to give more qualitative results .
figures 3 is similar to figure 1 , presenting the radial profiles of density , pressure , rotational velocity , and radial velocity . figure 4 shows the radial profiles of inflow and outflow rates and the net rate . from figure 3 , we see that the equatorial density , gas pressure , rotation velocity and radial velocity can again be well described by a power - law form .
the slopes of the profiles of rotation and radial velocity remain the same with model a. the normalization of the radial velocity is 10 times larger , as expected . compared to model a , the profiles of density and pressure are moderately steeper , while that of the inflow rate becomes flatter , ( r ) r^-0.85 , p ( r ) r^-1.85 , _
in(r)r^0.4.again more precisely , the inflow rate is described by _
in(r ) r^0.54[modelbmdotprofile]from @xmath7 to @xmath71 , while it is almost constant within @xmath8 .
the outflow rate is much steeper .
beyond @xmath79 it can be described by _
out(r)r^0.73.again we see that the relationship of @xmath76 is reasonably satisfied beyond @xmath8 , as in the case of @xmath4 .
we have also conducted simulations with @xmath88 .
we found that convective outflow becomes significantly weaker .
the density profile becomes steeper while the accretion rate profiles becomes flatter . in model
a and b , a rotating torus is adopted as the initial condition of our simulation .
we also study the cases of other initial condition .
one is to inject gas from an outer boundary , with the properties of the injected gas , including the temperature , rotation and radial velocity , being determined by the self - similar solution of an adaf ( narayan & yi 1994 ) .
another model is that the initial condition is expanded from the one dimensional global solution of adafs .
we find that the radial profiles of inflow rate and density are very similar ( paper ii ) .
however , this will not be the case if the angular momentum of the injected gas in the initial condition is very low . in that case
, the profile of accretion rate will become flatter and correspondingly density profile steeper .
the full discussion of the effect of initial condition and boundary condition will be presented in a subsequent paper ( bu , yuan & wu 2012 ) .
narayan & yi ( 1995 ; see also narayan , mahadevan & quataert 1998 ) presented the radial scaling of many quantities of the hot accretion flow based on the self - similar solution .
these solutions are very useful to estimate approximately the properties of hot accretion flow .
but at that time , the inward decrease of the accretion rate has not been found and taken into account in the solution .
for the convenience of future use , here we present the new scaling law . in terms of interpreting observations ,
the most prominent effect of outflow is on the profile of density .
therefore , compared to narayan & yi ( 1995 ) , only the scaling of density - related quantities such as the number density , magnetic field strength , pressure , and viscous dissipation rate , are changed .
the scaling of other quantities are also presented for completeness . in the original scaling of narayan & yi ( 1995 ) , the parameter @xmath89 ( @xmath90 , with @xmath91 .
here we do not specify a value for @xmath92 .
the new results are , @xmath93 @xmath94 @xmath95 @xmath96 @xmath97 @xmath98 @xmath99 _ es24 ^ -1_outr^1-pr_out^p-3/2same with narayan & yi ( 1995 ) , all quantities are written in scaled units , @xmath100 , @xmath101 is radius in unit of @xmath0 , @xmath102 ( @xmath103 , @xmath104 is the mass accretion rate at the outer boundary @xmath105 .
spb99 found that , for the @xmath20-description ( run k in spb99 ) as we adopte in the present work , @xmath106 .
igumenshchev & abramowicz ( 2000 ) performed two - dimensional hd simulations , considering a larger range in the parameter space spanned by @xmath20 and adiabatic index @xmath107 .
they found that when @xmath108 , for both @xmath85 and @xmath109 the density profile is roughly the same , i.e. , @xmath110 .
this result is approximately consistent with spb99 .
the density profile in these two works is flatter than our result since a newtonian potential was adopted there .
mckinney & gammie ( 2002 ) obtained @xmath111 and @xmath112 . but
this result strongly suffers from the `` inner boundary effect '' , since a pseudo - newtonian potential was adopted and the above scaling was measured over @xmath113 .
many mhd numerical simulations of hot accretion flow have been done after the pioneer hd works of igumenshchev & abramowicz ( 1999 , 2000 ) and spb99 . for two - dimensional simulations , since turbulence and accretion die away because there is no dynamo action to maintain the poloidal magnetic field , steady state can not be reached at large radii .
this effect , combined with the inner boundary effect , severely restricts the measurement of the radial profiles of dynamical quantities such as density . for three - dimensional simulation
, it is very time - consuming to simulate a large dynamical range . perhaps due to these reasons
, only a few works have presented radial profiles of density and inflow rate .
for example , stone & pringle ( 2001 ) , hawley , balbus & stone ( 2001 ) , and hawley & balbus ( 2002 ) did the two and three - dimensional mhd simulation of accretion flow .
they all found that the mass accretion rate decreases with decreasing radius , but the quantitative radial profile was not given .
machida , matsumoto & mineshige ( 2001 ) performed three - dimensional mhd simulation of hot accretion flows , with a toroidal initial magnetic field configuration .
a newtonian potential was adopted thus their results are not affected by the `` inner boundary effect '' .
they found that the radial structure of the accretion flow is very similar to that obtained by hd simulations .
two fits were obtained . in the `` early stage '' of the simulation they found that the radial profiles of the density and radial velocity can be described by @xmath60 and @xmath114 . at the `` late stage '' of their simulation
, they found moderately different results , @xmath115 and @xmath116 .
while the profile of density is very similar to our result , their profile of radial velocity is much steeper .
this may be because that the averaging approach adopted in their work is different from ours .
they average the quantity over the whole @xmath54 angles ( eqs .
[ 1 ] and [ 2 ] in machida , matsumoto & mineshige 2001 ) ; while our average is constrained to @xmath117 .
this approach is expected to produce a quite different result only to the radial velocity , since the radial velocity away from the equatorial plane is much larger than that close to @xmath118 while density is on the opposite .
igumenshchev , narayan & abramowicz ( 2003 ) also presented three - dimensional mhd simulation .
they adopted paczysky & wiita potential and two types of initial magnetic field configurations . only in the case of a toroidal initial configuration
the density profile was given , which is @xmath119 .
this profile is steeper than that of machida , matsumoto & mineshige ( 2001 ) .
the discrepancy should not because of the inner boundary effect , since the measurement is over @xmath120 .
we speculate that one reason may be that a different potential was adopted .
another reason may be associated with the strong fluctuation of the radial velocity in this radial range ( see fig .
11 in igumenshchev , narayan & abramowicz 2003 ) . in the above - mentioned mhd simulations ,
the magnetic field is weak .
recently attention has been paid to accretion flows with accumulated strong magnetic flux , i.e. , `` magnetically arrested disk '' or `` magnetically choked accretion flows '' ( narayan , igumenshchev & abramowicz 2003 ; pen et al .
2003 ; pang et al . 2011 ; mckinney , tchekhovskoy & blandford 2012 ) .
pen et al .
( 2003 ) constructed three - dimensional mhd simulation with huge number ( @xmath121 ) of grid zones .
they emphasized that the axial and reflection symmetries usually adopted in many simulations are not adopted in their work .
the input magnetic field is different from the usual way .
it it an admixture of random field loops and large loops that thread the whole simulation box .
the density profile they measured is @xmath122 over the radial range of @xmath123 , with @xmath124 is the bondi radius .
the large distance of the inner boundary from the black hole implies that this result does not suffer from the `` inner boundary effect '' .
this slope is well between the results of our model a and model b. in another three - dimensional mhd simulation by pang et al .
( 2011 ) , special attention was paid to the radial density ( or equivalently accretion rate ) profile of the accretion flow .
a newtonian potential was adopted in their work .
the outer boundary is quite large , almost ten times larger than the bondi radius .
they conducted a numerical survey of parameter space , namely the magnitude of the ambient magnetic field , the radial dynamical range within bondi radius , and the resolution of the bondi scale .
the found that the mass accretion rate at the inner boundary @xmath125 can be well described by the following power - law form , = ( ) ^3/2-k_eff , with the mostly favored value of @xmath126 .
here @xmath127 is the accretion rate at bondi radius @xmath128 and @xmath129 . again like the case of pen et al .
( 2003 ) , the rather large @xmath125 ensures that the obtained profile is not affected by the inner boundary effect .
this result is in good agreement with our model b ( eq.[[modelbmdotprofile ] ] ) .
mckinney , tchekhovskoy , & blandford ( 2012 ) simulated three - dimensional general relativistic mhd accretion flow with different thickness and two types of initial magnetic field configuration , namely poloidal and toroidal .
density profiles haven been obtained for `` thick '' flows , which is @xmath111 , same for both configurations .
for `` thin '' flows , it is @xmath130 .
however , the measurement is not relevant to us since it is over @xmath131 , where the inner boundary effect is very strong .
narayan et al .
( 2012 ) most recently presented two long - duration grmhd simulations of a hot accretion flow , in which the accumulation of magnetic flux around the black hole does and does not occur ( `` mad '' and `` sane '' models ) , respectively .
the steady state is reached up to @xmath132 . after taking into account the `` inner boundary effect '' and `` @xmath133 effect '' , they found @xmath62 when @xmath59 .
regarding the density profile , they claimed that its radial scaling is such that @xmath134 constant is roughly satisfied at small radii .
note that this is not in conflict with our results . as we have stated before and shown by figs .
2&4 , @xmath135 when @xmath136 .
outside @xmath7 the radial inflow rate profiles for the `` mad '' and `` sane '' models are @xmath137 and @xmath138 , respectively
( ramesh narayan , private communications ) .
the result of the `` sane '' model is in good consistency with our model a ( eq.[[inflowprofile2 ] ] ) while the result of `` mad '' model is similar to our model b ( eq.[[modelbmdotprofile ] ] ) . from the above - mentioned works we note that the inflow rate profile of accretion flow with accumulated magnetic flux seems to be somewhat flatter than that without accumulation .
this may be because that when there is accumulation of magnetic flux , the `` effective @xmath20 '' is larger , i.e. , `` mad '' and `` sane '' correspond to our model b and a , respectively .
overall , almost all current numerical simulations give a similar radial density profile , with @xmath139 .
the results depend weakly on the presence or absence of magnetic field ( hd or mhd ) , viscous parameter @xmath20 , strength of the magnetic field ( weak or strong ) , the initial configuration of the magnetic field ( toroidal or poloidal or tangled ) , and the dimension of calculation ( two or three dimension ) .
if this is true , an interesting question is then : what causes such a result ? to answer this question , one first needs to understand what causes the inward decrease of the mass accretion rate .
this question will be studied in paper ii .
here we briefly summarize the results .
two scenarios have been proposed to explain this result .
one is the convection - dominated accretion flow ( cdaf ) model ( narayan et al .
2000 ; quataert & gruzinov 2000a ) .
this model proposes that both hydro and mhd accretion flows are convectively unstable . in this case
the inward decrease of accretion rate is because that the fluid circulates in the convective eddies .
in contrast to the cdaf model , the adiabatic inflow - outflow solution ( adios ) model ( blandford & begelman 1999 , 2004 ; begelman 2012 ) suggests that the inward decrease of accretion rate is because of mass loss in the outflow .
our analysis in paper ii favors the latter .
moreover , we find that outflow is driven by buoyant force and magnetic centrifugal force in hd and mhd accretion flows , respectively . then it is an interesting question why different physical mechanisms cause the roughly same profiles of accretion rate and density ?
begelman ( 2012 ) gives an answer to this question . in the early version of the adios model ( blandford & begelman 1999 , 2004 )
, there are inflowing and outflowing zones with equal but opposite mass fluxes that vary with radius following a power - law form , @xmath140 . the net rate , which is the sum of inflow and outflow rates , is very small .
they applied the conservation laws of hydrodynamics to the inflow zone and obtained relationships among the parameters such as the specific angular momentum , effective binding energy , and the value of @xmath11 . in this model , the value of @xmath11
is not determined but only constrained to be in the range @xmath141 .
one advantage of the adios model , as claimed by the authors , is that although it was based on hd analysis , the approach can be fully applied to mhd accretion flows as well .
but the disadvantage is that the early adios model is intrinsically one - dimensional , the outflow zone is laminar and their streamlines are self - consistently patched onto the inflow zone at each radius . begelman ( 2012 ) recently developed this model .
the main improvement is that now the inflow and outflow zones are treated on an equal footing , and the outflow is permitted to be as turbulent as inflow rather than a laminar flow .
he found that the viable range of @xmath11 is now narrowed down from @xmath141 to @xmath142 .
the main reason for the regulation of the solution is an introduction of a conserved outward flux of energy through the flow , which was ignored in blandford & begelman ( 1999 , 2004 ) .
this work therefore well explains why various simulations obtain very similar profiles of accretion rate and density . because of this reason
, we have confidence that our hd simulation results should be similar to the more realistic mhd simulation results , and further can be compared to observations .
of course , we note that the quantitative result of the value of @xmath11 obtained in begelman ( 2012 ) is moderately larger than that obtained in simulations . in model
a , when @xmath143 where the `` inner boundary effect '' is the weakest thus more appropriate to be compared with begelman ( 2012 ) , we find @xmath144 . because of its proximity , the supermassive black hole located at the center of our galaxy is regarded as the unique laboratory for the study of black hole accretion ( see yuan 2011 for a review ) .
abundant of data has been obtained , which puts the most strict constrains to the theory of black hole accretion . among them
is the constrain on the mass accretion rate at both the bondi radius and the region close to the black hole horizon .
the high spacial resolution of _ chandra _ can well resolve the bondi radius and infer the gas density and temperature there ( baganoff et al .
the values of the bondi radius and the bondi accretion rate are then well determined , @xmath145 and @xmath146 , respectively .
although the value of mass accretion rate is obtained by the simple analytical bondi theory , it was found to be in good consistency with the detailed three - dimensional numerical simulations focusing on the fueling of the supermassive black hole in sgr a * of cuadra et al .
in fact , the numerical simulations obtained in this work is @xmath147 , only a factor of 3 lower than @xmath127 .
this factor is likely because of the inclusion of angular momentum of the accretion flow in the simulation . the accretion rate at the innermost region of the accretion flow , on the other hand , is strongly constrained by the radio polarization observations ( e.g. , quataert & gruzinov 2000b ; macquart et al .
2006 ; marrone et al .
the detected high level of linear polarization at frequencies higher than @xmath148 sets an upper limit to the rotation measure , which then requires that the accretion rate at the innermost region of the accretion flow must be in the range between @xmath149 and @xmath150 , depending on the assumed configuration of the magnetic field .
this implies that more than @xmath151 of the gas captured at the bondi radius must be lost and does not fall onto the black hole .
correspondingly , the density profile of the accretion flow is significantly flatter than the prediction of @xmath152 . the detailed modeling to sgr a * presented in yuan ,
quataert & narayan ( 2003 ) has shown that the radial profiles of accretion rate and density are described by @xmath153 and @xmath154 .
this density profile is moderately steeper than that obtained in our simulation .
this is likely because that the specific angular momentum of the accretion flow at the bondi radius is significantly sub - keplerian , @xmath155 times keplerian , as the numerical simulation has indicated ( cuadra et al .
2006 ) . in this case , our study ( bu , yuan & wu 2012 ) indicates that the outflow becomes significantly weaker , and the density profile correspondingly steeper .
ngc 3115 is an low - luminosity agn .
the mass of the black hole in its center is @xmath156 .
the source is very dim .
_ chandra _ observations only present an upper limit of @xmath157 ( wong et al .
2011 and references therein ) .
therefore the accretion mode in this source must be an adaf . at a distance of 9.7 mpc ,
ngc 3115 is the nearest @xmath158 black hole .
therefore , the angular size of the accretion flow in this source is very large , which is very helpful for us to directly detect the accretion flow .
wong et al . ( 2011 ) recently conducted _ chandra _ observation to this source and determined that the bondi radius is about @xmath159 .
the most prominent result is that for the first time , they have resolved the accretion flow within the bondi radius .
they found that the temperature is rising toward the black hole as expected in all accretion flow models . but more importantly
, the radial density profile of the accretion flow within @xmath160 is found to be well described by a power - law form , ( r ) r^-1.03^+0.23_-0.21 .
this result is similar to sgr a * and is in reasonable agreement with our numerical simulations .
one of the most important finding of both hd and mhd numerical simulations of hot accretion flow in the recent years is that the mass accretion rate decreases with decreasing radius .
correspondingly , the density profile becomes flatter compared to the previous prediction when the accretion rate is constant of radius .
one main problem with previous simulation is that because of technical difficulty the radial dynamical range in simulations is usually very limited , typically spanning less than two orders of magnitude .
the boundary condition effect makes the radial range over which we can reliably measure the profile of physical quantities even smaller .
the previous simulation results are therefore somewhat suspectable . in this paper
we adopt a `` two - zone '' approach to simulate an axisymmetric accretion flow , extending the dynamical range to over four orders of magnitude , i.e , from @xmath161 to @xmath162 .
we confirm previous results that the profiles of inflow rate and density can be well described by power - law forms . within @xmath8 , @xmath163 . beyond @xmath8 ,
the power - law slopes are a function of the viscous parameter @xmath20 . for @xmath4 , they are described by @xmath164 and @xmath165 . for @xmath85 ,
the results are @xmath166 and @xmath167 .
we also combine from literature all available numerical simulations which have presented the radial profile of density , both two - dimensional and three - dimensional , hd and mhd .
we find that all these simulations give somewhat similar results , @xmath168 , and this is also consistent with our results .
the diversity of the power - law index seems to come from the differences of the value of @xmath20 , the gravitational potential of the black hole , the initial condition of the simulation , and the strength of the magnetic field .
the rough consistency among various simulations can be explained by the most recent adios ( adiabatic inflow - outflow solution ) model ( begelman 2012 ) . in this work
it was found that after considering both the inflow and outflow zones at the equal footing and a conserved outward energy flux , the value of @xmath11 is well constrained to be in a narrow range .
the radial profiles of accretion rate and density obtained by numerical simulations are in good agreement with observations . in the case of sgr
a * , detailed modeling to the multi - waveband spectrum find that @xmath169 and @xmath154 . in the case of ngc 3115 , for the first time ,
_ chandra _ observations resolve the accretion flow within bondi radius , and find @xmath170 .
we are grateful to ramesh narayan for many valuable discussions and constructive comments .
we also thank jerry ostriker and jim stone for stimulating discussions on the two - zone approach adopted in this paper , and jon mckinney for the useful comments .
the hospitality of princeton university and copernicus astronomical center is acknowledged where part of the work was done .
this work was supported in part by the national basic research program of china ( 973 program 2009cb824800 ) , natural science foundation of china ( grants 10833002 , 10825314 , 11121062 , 11103059 , and 11133005 ) , and the cas / safea international partnership program for creative research teams .
the simulations were carried out at shanghai supercomputer center .
abramowicz , m. a. , chen x. , kato s. , lasota j .-
, regev o. , 1995 , apj,438 , l37 baganoff , f. k. , maeda y. , morris m. , bautz m. w. , brandt w. n. , cui w. , doty j. p. , feigelson e. d. , garmire g. p. , pravdo s. h. and 2 coauthors , 2003 , apj , 591 , 891 balbus , s. a. , hawley j. f. , 1991 , apj , 376 , 214 balbus , s. a. , hawley j. f. , 1998 , rev . mod .
, 70,1 blandford , r. d. & begelman , m. c. 1999 , mnras , 303 , l1 blandford , r. d. & begelman , m. c. 2004 , mnras , 349 , 68 begelman , m. c. 2012 , mnras , 420 , 2912 bu , d. f. , yuan , f. , & wu , m. c. 2012 , in preparation cuadra , j. , nayakshin , s. , springel , v. , & di matteo , t. 2006 , mnras , 366 , 358 de villiers , j. p. , hawley , j. f. , krolik , j. h. 2003 , apj , 599 , 1238 di matteo , t. , quataert , e. , allen , s. w. , narayan , r. , fabian , a. c. 2000 , mnras , 311 , 507 hawley , j. , balbus , s. a. 2002 , apj , 573 , 738 hawley , j. , balbus , s. a. , & stone , j. m. 2001 , apj , 554 , l49 hawley , j. f. , gammie , c. f. , & balbus , s. a. , 1996 , apj , 464 , 690 ho , l. , 2008 , ara&a , 46,475 ichimaru , s. 1977 , apj , 214 , 840 igumenshchev , i. v. , abramowicz m. a. , 1999 , mnras , 303,309 igumenshchev , i. v. , abramowicz m. a. , 2000 , apjs , 130,463 igumenshchev , i. v. , narayan , r. , & abramowicz , m. a. 2003 , apj , 592 , 1042 kato , s. , fukue j. & mineshige s. 1998 , black - hole accretion disks .
kyoto univ . press ,
kyoto li , j. , ostriker , j. p. , & sunyaev , r. 2012 , apj , in press ( arxiv:1206.4059 ) machida , m. matsumoto , r. , & mineshige , s. 2001 , pasj , 53 , l1 macquart , j. p. , et al .
2006 , apj , 646 , l111 marrone , d. p. , moran j. m. , zhao j. h. , rao r. , 2007 , apj , 654 , l57 mckinney , j. & gammie , c. f. 2002 , apj , 573 , 728 mckinney , j. , tchekhovskoy , a. , & blandford , r. 2012 , submitted to mnras ( arxiv:1201.4163 ) mushotzky , r. et al .
2000 , aas high energy astrophysics division , 32 , 2101 narayan , r. 2005 , ap&ss , 300 , 177 narayan , r. , igumenshchev , i. v. , & abramowicz , m. a. 2000 , apj , 593 , 798 narayan , r. , igumenshchev , i. v. , & abramowicz , m. a. 2003 , pasj , 55 , l69 narayan , r. , mcclintock j.e .
, 2008 , new astron .
, 51 , 733 narayan , r. , & yi , i. 1994 , apj , 428 , l13 narayan , r. , & yi , i. 1995 , apj , 452 , 710 narayan , r. , mahadevan , r. , & quataert , e. 1998 , in abramowicz m. a , bjornsson g. , pringle j. e. eds , the theory of black hole accretion discs .
cambridge univ . press , cambridge narayan , r. , sadowski , a. , penna , r. f. et al .
2012 , submitted to mnras ( arxiv:1206.1213 ) paczysky , b. & wiita , p. j. 1980 , a&a , 88 , 23 pang , b. et al .
2011 , mnras , 415 , 1228 papaloizou , j. c. b. , pringle j. e.,1984 mnras , 208,721 pen , u .- l . ,
matzener , c.d . , &
wong , s. 2003 , apj , 596 , l207 proga , d. , & begelman , m. 2003a , , 582 , 69 proga , d. , & begelman , m. 2003b , , 592 , 767 quataert , e. , & gruzinov a. , 2000a , apj , 539 , 809 quataert , e. , & gruzinov a. , 2000b , apj , 545 , 842 quataert , e. , & narayan , r. 1999 , apj , 520 , 298 rees , m. j. , begelman , m. c. , blandford , r. d. , & phinney , e. s. 1982 , nat , 295 , 17 stone , j. m. , hawley , j. f. , gammie , c. f. , balbus , s. a. , 1996 , apj , 463 , 656 stone , j. m. , & norman , m. l. 1992 , apjs , 80 , 753 stone , j. m. , & pringle , j. e. 2001 , mnras , 322 , 461 stone , j. m. , pringle , j. e. , & begelman , m. c. 1999 , mnras , 310 , 1002 ( spb99 ) wong , k. w. , et al .
2011 , apj , 736 , l23 yuan , f. 2007 , asp conf .
373 , the central engine of active galactic nuclei , ed .
l. c. ho and j .- m .
wang , 95 yuan , f. 2011 , asp conf .
439 , the galactic center : a window to the nuclear environment of disk galaxies , ed .
mark r. morris , q. daniel wang , and feng yuan , 346 yuan , f. & bu , d. 2010 , mnras , 408 , 1051 yuan , f. , bu , d. , & wu , m. 2012 , apj , in press ( arxiv : 1206.4173 ) ( paper ii ) yuan , f. & narayan , r. 2013 , ara&a , in preparation yuan , f. , quataert , e. & narayan , r. 2003 , apj , 598 , 301 | numerical simulations of hot accretion flow , both hydrodynamical and magnetohydrodynamical , have shown that the mass accretion rate decreases with decreasing radius ; consequently the density profile of accretion flow becomes flatter compared to the case of a constant accretion rate .
this result has important theoretical and observational implications .
however , because of technical difficulties , the radial dynamic range in almost all previous simulations usually spans at most two orders of magnitude .
this small dynamical range , combined with the effects of boundary conditions , makes the simulation results suspectable .
especially , the radial profiles of density and inflow rate may not be precise enough to be used to compare with observations . in this paper
we present a `` two - zone '' approach to expand the radial dynamical range from two to four orders of magnitude .
we confirm previous results and find that from @xmath0 to @xmath1 the radial profiles of accretion rate and density can be well described by @xmath2 and @xmath3 .
the values of ( s , p ) are ( 0.48 , 0.65 ) and ( 0.4 , 0.85 ) , for viscous parameter @xmath4 and @xmath5 , respectively . or more precisely , the accretion rate is constant ( i.e. , @xmath6 ) within @xmath7 ; but beyond @xmath8 , we have @xmath9 and @xmath10 for @xmath4 and @xmath5 , respectively . we find that the values of both @xmath11 and @xmath12 are similar in all numerical simulation works , including previous and the present ones , no matter a magnetic field is included or not and what kind of initial conditions are adopted
. such an apparently surprising `` common '' result can be explained by the most updated version of the adiabatic inflow - outflow model ( adios ) .
the density profile we obtain is in good quantitative agreement with that obtained from the detailed observations and modeling to sgr a * and ngc 3115 .
the origin and implication of such a profile will be investigated in a subsequent paper . |
the charmonium singlet states have resisted a firm identification for many years @xcite , but , now the @xmath2 , @xmath3 and @xmath4 are well identified .
the mass of the latter , as given by the cleoc collaboration is @xcite @xmath5 , where the first uncertainty comes from statistics and the second one from estimated systematic errors . for the previous attempts and other measurements , see , e.g. , @xcite .
it is perfectly legitimate to _ define _ the hyperfine splitting by @xmath6 .
this leads to the experimental value @xcite @xmath7 . but a superficial reading of @xmath8 could be misleading , as it suggests a very small or even vanishing effect of spin spin forces in the 1p multiplet , while it is arguably larger , and positive , of the order of a few mev . in the potential models ,
if the spin - orbit , @xmath9 and tensor , @xmath10 terms are treated in perturbation , the masses of triplet p - states with @xmath11 , @xmath12 and @xmath13 are shifted by @xmath14 and @xmath15 , respectively , and it is readily seen that the contributions of @xmath16 and @xmath17 disappear in the @xmath18-weighted centre of gravity .
now the @xmath19 splitting is of the order of 150mev , suggesting that at the level of 1mev accuracy , the calculation of the spin splittings should be pushed beyond first order . as the spin operators @xmath20 and
@xmath21 enter the hamiltonian linearly , the ground state of the spin - triplet p - state is a concave function of the values of these operators .
this means that in any potential model where the components @xmath22 and @xmath23 are suitably regularised and inserted non - pertubatively into the wave equation , the genuine spin - orbit- and tensor - free triplet state is _ above _ the naive centre of gravity . as an example , if one adopts the potential of ref .
@xcite , which is rather conventional in the heavy - quark sector ( the light - quark one is more speculative , with goldstone - boson exchanges ) , the difference is about 3mev , which locates the experimental @xmath4 about 4mev above this improved triplet benchmark .
unfortunately , the convexity effect depends rather sensitively on the details of the regularisation of the spin - orbit and tensor terms , and many other effects have to be taken into account , for instance , the p f orbital mixing , which pushes down the @xmath24 state .
phenomenological potentials could also include further spin operators that are not necessary in the simplest non - relativistic reduction of one - gluon exchange , or in the thomas precession , and , for spin triplet , can not be reduced to spin orbit and tensor and thus do not average out to zero in the naive centre of gravity .
an example is the `` quadratic spin orbit '' operator used in nuclear physics to describe the nucleon nucleon potential . on the theoretical side
, the zero - range property of the spin - spin interaction , @xmath25 only holds in the simplest non - relativistic reduction la breit fermi of the coulomb term due to one - gluon - exchange . a range of about the inverse quark mass would be reasonable , and would give a non - vanishing matrix element for p - states .
higher - order terms in perturbative qcd have been discussed in the literature , see , e.g.,@xcite and refs . there .
the spin spin potential has also been estimated non - perturbatively with the lattice techniques @xcite , exhibiting a range that is short but finite . in some other lattice studies
@xcite , the states @xmath26 and @xmath4 are calculated directly ( not la born oppenheimer through a potential ) , each with a specific operator adapted to its quantum numbers , as in qcd sum rules @xcite .
the interpretation of the mass of the @xmath27 also reveals the limits of simple potentials .
the splitting @xmath28 , is appreciably smaller than the predicted one in most @xmath29 constituent models tuned to reproduce @xmath30mev .
a likely explanation is that for this 2s multiplet lying very close to the @xmath31 threshold , the coupling to higher fock configurations is enhanced .
the @xmath32 is pushed down by the very close @xmath31 threshold , while @xmath27 is less affected , since only influenced by the higher lying @xmath33 and @xmath34 thresholds .
this reduces the effect of the quark
antiquark spin spin forces @xcite . hopefully , the @xmath35 will be found shortly .
its mass can be estimated with sophisticated techniques . if one accounts for the @xmath36 factor in front of the spin - spin interaction , and the squeezing of wave - functions when the constituent masses increase , which for a logarithmic potential gives a factor @xmath37 for the squared wave function at the origin , one gets @xmath38 in ratio @xmath39 to its charm analogue , i.e. , about 68mev .
a few years ago , several new results came in the meson sector with single charm , in particular the @xmath40 states , and this stimulated an intense activity . more recently , several new baryons have been found , and nowadays , the family of charmed baryons include many states @xcite .
a key question in baryon spectroscopy is to find evidence for the three - body structure , i.e. , states in which both @xmath41 and @xmath42 degrees of freedom are excited . in the harmonic - oscillator of light quarks , with flavour su(3 ) symmery , this corresponds to the @xmath43 multiplet , with an antisymmetric orbital wave - function @xmath44 $ ] that couples to an antisymmetric spin
isospin wave function and an antisymmetric colour wave function .
the lack of firm experimental candidate is perhaps due to the small cross - sections in pion- or photo - production experiments , which favour states having better overlap with the quark wave function of the target nucleon .
another picture is proposed by diquark models , in which these states do not exist , if the diquark is in its ground state .
perhaps the first baryon with excitation in both jacobi variables will be found in the charm sector : this state is expected to be rather narrow and to have preferentially at least one orbital excitation in its decay products . among ordinary hadrons ,
the @xmath45 baryons with two heavy quarks are particularly interesting , as they combine the adiabatic motion of two heavy quarks , as in charmonium , and the relativistic motion of a light quark around a coloured source , as in @xmath46 mesons .
the ground state has interesting weak - decay properties.there are huge differences among the lifetimes of @xmath46 , @xmath47 mesons and single - charm baryons .
the hierarchy of the lifetimes is well understood in terms of @xmath48-exchange , or interference effects , but the differences are usually larger than estimated in calculations . in the case of hadrons with two heavy quarks , binding effects also play a role .
selex has serious candidates for the ground states , and more fragile evidence for the isospin partner , spin or orbital excitations @xcite .
however , other experiments were not able to find any double - charm baryons , in particular in @xmath49 @xcite .
this is a little surprising , because meanwhile the b - factories found an excess ( vs. simple qcd expectations ) of double charm - pair production , @xmath50 , leading to beautiful @xmath29 spectra recoiling against the j/@xmath51 .
one would naively expect that if a @xmath52 primary system is easily produced , it sometimes rearranges into a doubly - charm diquark and a conjugate antidiquark .
when the first studies of double - charm baryons were carried out , their experimental study was out of reach .
now , with the @xmath53 well measured , and the first indication for @xmath54 , the sector of two heavy flavours seems ready for detailed spectroscopy , and one could already envisage one step beyond , i.e. , triple charm . the spectrum of @xmath55 was called @xcite `` the ultimate goal of baryon spectroscopy '' , the true analogue of charmonium for baryons . here
the three - quark dynamics can be tested in the static limit and confronted with theory .
for instance , the level order is expected to be similar to that of charmonium , with the first excitation having a parity opposite to that of the ground - state .
remember that for light baryons , the roper resonance , with the same positive parity as the ground - state nucleon , comes slightly lower than the first orbital excitations , and this can not be accommodated in simple quark models .
several intriguing states have been identified in the hidden - charm sector , that are hardly compatible with genuine @xmath29 states . the experimental situation concerning the @xmath56 , the various @xmath0 s near 3940 , the @xmath57 , etc .
, is reviewed in several contributions to this conference @xcite .
the most popular explanation of @xmath56 is that of a @xmath58 molecule , see , e.g. , @xcite and refs .
there on the pioneering works by voloshin et al . ,
trnqvist , glashow et al . , ericson and karl , manohar et al . , braaten et al .
nuclear forces acting between charmed mesons generate a nuclear potential which is weaker that the proton neutron spin - triplet interaction , but being experienced by heavier particles , it gives comparable spectral properties , at the edge between binding and non - binding .
interesting developments have been proposed , in particular bound sates of two or several charmed or doubly - charmed baryons @xcite .
also , as the @xmath46 and the @xmath59 are not strictly bound in this approach , but slightly above their threshold , one could envisage the borromean binding of three or more heavy mesons .
there are , however , some caution in order . as stressed by suzuki @xcite , due to the @xmath60 mass difference , the yukawa potential in @xmath58 is non local , and this might weaken its efficiency .
also , the miracle in nuclear physics is the presence of a hard core , which prevents the nucleons from collapse and reinforces the role of long - range dynamics . there is no such hard core in @xmath58 , and one should account for the direct interaction between the quarks of @xmath46 and these of @xmath59 .
this leads us to the alternative four - quark models . among these models
, there is the diquark antidiquark picture , as developed in particular by maiani et al .
@xcite , giving an unified picture of several new states .
notice that the diquark is an effective cluster , an approximation valid only in a given environment .
if taken too seriously , some of the diquark models of @xmath0 and @xmath1 , with a relatively low mass for the diquarks , could lead to predict the existence of stable triple - charm dibaryons , such as @xmath61 , below the @xmath62 threshold .
the four - quark dynamics , and its application to @xmath56 , is also discussed by lipkin and hgaasen et al .
@xcite , among others .
the chromomagnetic interaction , with a realistic treatment of flavour - symmetry breaking gives a simple explanation of the mass and decay properties of @xmath56 @xcite .
now , a lesson from atomic physics , is that the best place for stable four - body states is not @xmath63 , which is slightly stable for @xmath64 , but loses stability for @xmath65 @xcite .
however , the configurations @xmath66 are more stable that these with equal masses @xcite .
the crucial rule is that the coulomb interaction remains unchanged when the masses evolve from electron to muon or heavier constituents . in qcd
, we have the same property , called _ flavour independence _ , for the spin - independent interaction .
this is why states of the type @xmath67 are predicted to exist @xcite , at least in the limit of large @xmath68 mass ratio .
their production and identification could be carried out in the experiments searching for the double - charm baryons .
a good surprise of recent high - energy experiments has been the ability of performing productively in hadron physics , and even to clarify the results claimed by dedicated low - energy experiments .
another good surprise is the ability to produce fragile and complex structures , such as antideuterium @xcite , bound only by 2mev .
it is reasonable to anticipate significant progress on heavy hadrons , in particular exotic multiquarks , from the future lhc experiments , provided a small fraction of the analysis is devoted to this physics .
99 a. martin and j. m. richard , cern cour .
* 43n3 * ( 2003 ) 17 ; t. barnes , t. e. browder and s. f. tuan , arxiv : hep - ph/0408081 .
p. rubin _ et al .
_ [ cleo collaboration ] , phys . rev .
d * 72 * ( 2005 ) 092004 .
m. andreotti _ et al .
_ , phys .
d * 72 * ( 2005 ) 032001 .
see several contributions at this dis conference , in particular by k. seth , v. poireau , g. pakhlova and s. chekanov , ` http://www.mppmu.mpg.de/dis2007/ ` .
j. vijande , f. fernandez and a. valcarce , j. phys .
g * 31 * ( 2005 ) 481 .
n. brambilla _ et al . _
[ qwg coll .
] , cern yellow report 2005 - 005 [ arxiv : hep - ph/0412158 ] . s. recksiegel and y. sumino , phys .
b * 578 * ( 2004 ) 369 .
s. narison , `` qcd as a theory of hadrons '' , cambridge univ .
press , ( cambridge , u.k . , 2004 ) .
a. martin and j. m. richard , phys .
b * 115 * ( 1982 ) 323 ; e. j. eichten , k. lane and c. quigg , phys .
d * 73 * ( 2006 ) 014014 [ erratum - ibid .
d * 73 * ( 2006 ) 079903 ] ; e. eichten , s. godfrey , h. mahlke and j. l. rosner , arxiv : hep - ph/0701208 ; m. r. pennington and d. j. wilson , arxiv:0704.3384 [ hep - ph ]
. j. engelfried [ selex collaboration ] , _ in the proceedings of international conference on heavy quarks and leptons ( hql 06 ) , munich , germany , 16 - 20 oct 2006 , pp 003 _ [ arxiv : hep - ex/0702001 ] .
b. aubert _ et al .
_ [ babar collaboration ] , phys .
d * 74 * ( 2006 ) 011103 .
bjorken , conf .
hadron spectroscopy ( college park , 1985 ) ed .
s. oneda ( alp , new york , 1985 ) .
e. s. swanson , phys .
* 429 * ( 2006 ) 243 .
f. froemel , b. julia - diaz and d. o. riska , nucl .
a * 750 * ( 2005 ) 337 .
m. suzuki , phys .
d * 72 * ( 2005 ) 114013 .
l. maiani , v. riquer , f. piccinini and a. d. polosa , phys .
d * 72 * ( 2005 ) 031502 , and refs . therein .
h. hogaasen , j. m. richard and p. sorba , phys .
d * 73 * ( 2006 ) 054013 ; f. buccella et al .
j. c * 49 * ( 2007 ) 743 ; h. j. lipkin , arxiv : hep - ph/0703190 .
armour et al . , phys . rep .
* 413 * ( 2005 ) 1 .
j. p. ader , j. m. richard and p. taxil , phys .
d * 25 * ( 1982 ) 2370 ; d. m. brink and f. stancu , phys .
d * 57 * ( 1998 ) 6778 ; b. a. gelman and s. nussinov , phys .
b * 551 * ( 2003 ) 296 ; d. janc and m. rosina , few body syst .
* 35 * ( 2004 ) 175 . | caution is suggested on the comparison of the spin - singlet charmonium p - state with the centre of gravity of triplet states , when the mass splitting is of the order of a few mev .
the physics of new hidden - charm states @xmath0 and @xmath1 is briefly reviewed .
perspectives for producing double - charm baryons and double - charm exotic mesons are discussed . |
this paper describes the background for topological quantum computing in terms of temperely
lieb recoupling theory .
this is a recoupling theory that generalizes standard angular momentum recoupling theory , generalizes the penrose theory of spin networks and is inherently topological .
lieb recoupling theory is based on the bracket polynomial model for the jones polynomial .
it is built in terms of diagrammatic combinatorial topology .
the same structure can be explained in terms of the @xmath1 quantum group , and has relationships with functional integration and witten s approach to topological quantum field theory .
nevertheless , the approach given here will be unrelentingly elementary .
elementary , does not necessarily mean simple .
in this case an architecture is built from simple beginnings and this archictecture and its recoupling language can be applied to many things including : colored jones polynomials , witten reshetikhin
turaev invariants of three manifolds , topological quantum field theory and quantum computing . in quantum computing ,
the application is most interesting because the recoupling theory yields representations of the artin braid group into unitary groups @xmath2 .
these represententations are _ dense _ in the unitary group , and can be used to model quantum computation universally in terms of representations of the braid group .
hence the term : topological quantum computation . in this paper
, we outline the basics of the temperely
lieb recoupling theory , and show explicitly how unitary representations of the braid group arise from it .
we will return to this subject in more detail in subsequent papers .
in particular , we do not describe the context of anyonic models for quantum computation in this paper .
rather , we concentrate here on showing how naturally unitary representations of the braid group arise in the context of the temperely lieb theory . for the reader interested in the relevant background in anyonic topological quantum computing
we recommend the following references \ { @xcite } .
here is a very condensed presentation of how unitary representations of the braid group are constructed via topological quantum field theoretic methods . for simplicity assmue
that one has a single ( mathematical ) particle with label @xmath3 that can interact with itself to produce either itself labeled @xmath4 or itself with the null label @xmath5 when @xmath6 interacts with @xmath3 the result is always @xmath7 when @xmath6 interacts with @xmath6 the result is always @xmath5 one considers process spaces where a row of particles labeled @xmath3 can successively interact , subject to the restriction that the end result is @xmath8 for example the space @xmath9 $ ] denotes the space of interactions of three particles labeled @xmath8 the particles are placed in the positions @xmath10 thus we begin with @xmath11 in a typical sequence of interactions , the first two @xmath3 s interact to produce a @xmath12 and the @xmath6 interacts with @xmath3 to produce @xmath8 @xmath13 in another possibility , the first two @xmath3 s interact to produce a @xmath14 and the @xmath3 interacts with @xmath3 to produce @xmath8 @xmath15 it follows from this analysis that the space of linear combinations of processes @xmath9 $ ] is two dimensional .
the two processes we have just described can be taken to be the the qubit basis for this space .
one obtains a representation of the three strand artin braid group on @xmath9 $ ] by assigning appropriate phase changes to each of the generating processes .
one can think of these phases as corresponding to the interchange of the particles labeled @xmath16 and @xmath17 in the association @xmath18 the other operator for this representation corresponds to the interchange of @xmath17 and @xmath19 this interchange is accomplished by a _
unitary change of basis mapping _
@xmath20 \longrightarrow v[a(bc)].\ ] ] if @xmath21 \longrightarrow v[(ba)c]\ ] ] is the first braiding operator ( corresponding to an interchange of the first two particles in the association ) then the second operator @xmath22 \longrightarrow v[(ac)b]\ ] ] is accomplished via the formula @xmath23 where the @xmath24 in this formula acts in the second vector space @xmath25 $ ] to apply the phases for the interchange of @xmath17 and @xmath19 in this scheme , vector spaces corresponding to associated strings of particle interactions are interrelated by _ recoupling transformations _ that generalize the mapping @xmath26 indicated above .
a full representation of the artin braid group on each space is defined in terms of the local intechange phase gates and the recoupling transfomations .
these gates and transformations have to satisfy a number of identities in order to produce a well - defined representation of the braid group .
these identities were discovered originally in relation to topological quantum field theory . in our approach
the structure of phase gates and recoupling transformations arise naturally from the structure of the bracket model for the jones polynomial @xcite .
thus we obtain a knot - theoretic basis for topological quantum computing .
in this section we discuss a combinatorial construction for spin networks that generalizes the original construction of roger penrose @xcite .
the result of this generalization is a structure that satisfies all the properties of a graphical @xmath27 as described in our paper on braiding and universal quantum gates @xcite , and specializes to classical angular momentum recoupling theory in the limit of its basic variable .
the construction is based on the properties of the bracket polynomial @xcite .
a complete description of this theory can be found in the book temperley
lieb recoupling theory and invariants of three - manifolds " by kauffman and lins @xcite .
the @xmath0-deformed " spin networks that we construct here are based on the bracket polynomial relation .
view figure 1 and figure 2 . [ cols="^ " , ] * theorem . * in the temperley lieb theory
we obtain unitary ( in fact real orthogonal ) recoupling transformations when the bracket variable @xmath24 has the form @xmath28 .
thus we obtain families of unitary representations of the artin braid group from the recoupling theory at these roots of unity .
* proof . *
the proof is given the discussion above . * remark . * the density of these representations will be taken up in a subsequent paper .
most of this effort was sponsored by the defense advanced research projects agency ( darpa ) and air force research laboratory , air force materiel command , usaf , under agreement f30602 - 01 - 2 - 05022 .
some of this effort was also sponsored by the national institute for standards and technology ( nist ) .
government is authorized to reproduce and distribute reprints for government purposes notwithstanding any copyright annotations thereon .
the views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements , either expressed or implied , of the defense advanced research projects agency , the air force research laboratory , or the u.s .
( copyright 2006 . ) | we review the @xmath0-deformed spin network approach to topological quantum field theory and apply these methods to produce unitary representations of the braid groups that are dense in the unitary groups . |
the statistical properties of fully developed , homogeneous , isotropic turbulence are often characterised by velocity structure functions , which are averages of the differences of fluid velocities at two points separated by a distance @xmath0 ( a precise definition is given below ) .
if @xmath0 lies in the _ inertial range _ of scales that lie between the large length scale @xmath1 , at which energy is pumped into a turbulent fluid , and the small dissipation scale @xmath2 , at which viscous dissipation becomes significant , these structure functions scale as a power of @xmath0 , in a manner that is reminiscent of the algebraic behaviour of correlation functions at a critical point in , say , a spin system .
this similarity between the statistical properties of homogeneous , isotropic turbulence and a system at a critical point is well known ; and its elucidation for _ equal - time _ structure functions has been the subject of many papers : it turns out that the simple scaling we are accustomed to at most critical points must be generalised to multiscaling in turbulence ; i.e. , an infinity of exponents is required to characterise the inertial - range behaviours of structure functions of different orders @xcite .
the scaling behaviour of correlation functions at a critical point @xcite is associated with the divergence of a correlation length @xmath3 at the critical point ; e.g. , in a spin system @xmath4 if the field @xmath5 , where the reduced temperature @xmath6 , @xmath7 is the temperature , @xmath8 the critical temperature , and @xmath9 is an equal - time critical exponent that is universal ( for systems in a given universality class ) . in the vicinity of a critical point the relaxation time @xmath10 , which can be determined from _ time - dependent _ correlation functions , scales as follows : @xmath11 this is known as the _ dynamic - scaling ansatz _ via which we define @xmath12 , the dynamic - scaling exponent . over the past few years
there has been considerable progress @xcite in developing the analogue of such a dynamic - scaling _ ansatz _ for _ time - dependent _ structure functions in homogeneous , isotropic turbulence .
we give an overview of this work on _ dynamic mutiscaling _ in turbulence .
we show , in particular , that ( a ) an infinity of dynamic - multiscaling exponents is required here , ( b ) these exponents depend on the precise way in which relaxation times are extracted from time - dependent structure functions , and ( c ) that dynamic - multiscaling exponents are related by _ linear bridge relations _ to equal - time multiscaling exponents .
the remaining part of this paper is organised as follows : section 2 contains an introduction to the multiscaling of equal - time structure functions in fluid turbulence .
section 3 is devoted to the dynamic multiscaling of time - dependent structure functions . in sect .
4 we introduce the goy shell model @xcite for fluid turbulence and give the details of our numerical studies of this model . section 5 gives representative results from our simulations for time - dependent structure functions for the goy shell model with conventional viscosity ; we also present new results with hyperviscosity and show explicitly that dynamic multiscaling exponents are independent of the type of viscosity we use .
we end with concluding remarks about the possibility of experimental verifications of our predictions and generalizations to other types of turbulence such as passive - scalar turbulence@xcite .
fluid flows are described by the navier - stokes equation for the velocity field @xmath13 at point @xmath14 and time @xmath15 : @xmath16 we consider low - mach number flows that are nearly incompressible and so equation ( [ ns1 ] ) must be augmented by the incompressibility constraint @xmath17 which can be used to eliminate the pressure @xmath18 in equation ( [ ns1 ] ) ; we choose the uniform density @xmath19 ; @xmath20 is the kinematic viscosity ; and for decaying turbulence the external force @xmath21 is zero .
turbulence occurs when the reynolds number @xmath22 is large ; here @xmath23 and @xmath24 are characteristic length and velocity scales of the flow .
the chaotic nature of turbulence has led to a natural statistical description of the velocity field in terms of velocity structure functions .
for instance , we can define the order-@xmath25 , equal - time , structure function for longitudinal velocity increments @xmath26.{\bf r}/r$ ] as follows : @xmath27^p \ra \sim r^{\zeta_p } ; \label{sp}\end{aligned}\ ] ] the power - law dependence on @xmath0 , which defines the exponent @xmath28 , holds for @xmath29 ; and the angular brackets indicate an average over either the statistical steady state , for forced turbulence , or statistically independent initial conditions , for decaying turbulence .
in contrast with the scaling behaviour of correlation functions at conventional critical points in equilibrium statistical mechanics , in turbulence the structure functions @xmath30 do not exhibit simple scaling forms : experimental and numerical evidence suggests that @xmath30 show multiscaling , with @xmath28 a nonlinear , convex , monotone increasing function of @xmath25 @xcite .
the 1941 theory of kolmogorov ( k41 ) @xcite yields simple scaling with @xmath31 ; the measured values of @xmath28 deviate significantly from @xmath32 for @xmath33 ; and for @xmath34 we have the exact von krmn - howarth @xcite result @xmath35 . even though the k41 phenomenology fails to capture the multiscaling of the equal - time structure function @xmath30 , it provides us with important conceptual underpinnings for studies of homogeneous , isotropic turbulence .
in particular , the exponents @xmath32 are _ universal _ in the sense that they do not depend on the details of the dissipation , i.e. , the viscosity .
( of course measurements should be made far away from boundaries so that the flow satisfies the conditions of homogeneity and isotropy . )
this suggestion of power - law scaling with universal scaling exponents was made for turbulence a few decades before it was appreciated fully in the context of critical phenomena .
the universality of the exponents @xmath28 holds even if we go beyond k41 phenomenology .
let us first examine the dependence of these exponents on the dissipation mechanism .
in numerical simulations it is possible to introduce a hyperviscosity @xmath36 by replacing the viscous term @xmath37 in equation ( [ ns1 ] ) by @xmath38 ; here @xmath39 determines the _ degree _ of hyperviscosity ; and @xmath40 yields normal viscous dissipation .
some early shell - model studies @xcite suggested that the exponents @xmath28 depend on @xmath41 .
but subsequent direct numerical simulations ( dns ) @xcite of equations ( [ ns1]-[ns2 ] ) and numerical studies of a shell model @xcite for turbulence have argued against this .
our results are consistent with these later studies .
the exponents @xmath28 do not seem to depend on whether they are measured in statistically steady , forced turbulence or in decaying turbulence .
numerical evidence for this _ universality _ has been provided by the shell - model studies of ref
. @xcite .
in the remaining sections of this paper we will show how to generalise the dynamic - scaling _ ansatz _ ( [ ds ] ) to account for multiscaling in turbulence .
our discussion will be based on earlier studies @xcite and the work carried out in our group @xcite .
we will then examine the universality of the dynamic - multiscaling exponents in fluid turbulence , i.e. , we will explore their dependence on ( a ) the hyperviscosity parameter @xmath41 and ( b ) the type of turbulence ( statistically steady as opposed to decaying ) .
the dynamic scaling of time - dependent correlation functions at a critical point in an equilibrium system was systematised soon after the scaling of equal - time correlation functions . the analogous development for homogeneous ,
isotropic turbulence has been carried out recently @xcite .
we summarise the essential points here before presenting our new results . a nave extension of k41 phenomenology to time - dependent structure functions
yields a dynamic exponent @xmath42 for all orders @xmath25 .
two improvements are required to go beyond this k41 result : ( a ) we must account for the multiscaling of velocity structure functions .
( b ) we must distinguish between structure functions of eulerian and lagrangian velocities ; the former yield trivial dynamic scaling with @xmath43 , for all @xmath25 , since the mean flow ( or the flow of the largest eddy ) directly advects small eddies and so temporal and spatial separations are related linearly by the mean flow velocity ; nontrivial dynamic multiscaling can be anticipated , therefore , only for lagrangian @xcite or quasi - lagrangian @xcite velocity structure functions . the latter
are defined in terms of the quasi - lagrangian velocity @xmath44 that is related to its eulerian counterpart @xmath45 as follows : @xmath46 , \label{qltrans}\ ] ] with @xmath47 the position at time @xmath15 of a largrangian particle that was at @xmath48 at time @xmath49 .
equal - time , quasi - lagrangian velocity structure functions are the same as their eulerian counterparts @xcite .
the order-@xmath25 , time - dependent , structure function , for longitudinal , quasi - lagrangian velocity increments is @xcite @xmath50 \ra .
\label{fp}\end{aligned}\ ] ] since we are interested in scaling behaviours , we restrict @xmath0 to the inertial range , consider , for simplicity , @xmath51 and @xmath52 , and denote the structure function ( [ fp ] ) by @xmath53
. given @xmath53 , there are different ways of extracting time scales .
for example , we can use the following order-@xmath25 , degree-@xmath54 integral- and derivative - time scales that are defined , respectively , as @xcite @xmath55^{(1/m ) } \label{timp}\ ] ] and @xmath56^{(-1/m)}. \label{tdp}\ ] ] if the integral in ( [ timp ] ) and the derivative in ( [ tdp ] ) exist we can generalise the dynamic - scaling _ ansatz _ ( [ ds ] ) at a critical point to the following dynamic - multiscaling _ anstze _ for homogeneous , isotropic turbulence @xcite : @xmath57 @xmath58 these equations define , respectively , the integral- and the derivative - time multiscaling exponents @xmath59 and @xmath60 , which satisfy the bridge relations @xmath61/m , \label{zipm}\ ] ] obtained first in ref .
@xcite , and @xmath62/m , \label{zdpm}\ ] ] obtained in ref .
these bridge relations follow from a generalisation @xcite of the multifractal formalism @xcite for turbulence . here
we sketch the arguments that lead to equation ( [ zipm ] ) for statistically steady turbulence and refer the reader to refs .
@xcite for details : we begin by assuming the following multifractal form for the order-@xmath25 time - dependent structure functions : @xmath63 as in the equal - time multifractal formalism @xcite , the scaling exponents @xmath64 $ ] . corresponding to each exponent @xmath65 there is a set @xmath66 , of fractal dimension @xmath67 and with a measure @xmath68 , such that @xmath69 if @xmath70 , with @xmath71 is such that @xmath72 and that the characteristic decay time @xmath73
. if we substitute for @xmath74 in equation ( [ zi ] ) , do the time integral first by a saddle - point method , we obtain the bridge relation ( [ zipm ] ) .
similar calculations lead to ( [ zdpm ] ) .
a complete discussion of such bridge relations , as well as similar relations for passive - scalar turbulence , can be found in refs .
the last of these references also shows that the bridge relations are the same for decaying and statistically steady turbulence .
given current computational resources , it has not been possible to verify the bridge relations ( [ zipm]-[zdpm ] ) by computing quasi - lagrangian velocities in a direct numerical simulation of the navier - stokes equation ( [ ns1]-[ns2 ] ) for an incompressible fluid .
thus we must turn to numerical studies of simple shell models of turbulence that we describe in the next section .
we have carried out extensive numerical simulations to obtain equal - time and dynamic multiscaling exponents for the goy shell model for fluid turbulence @xcite .
this shell model is defined on a logarithmically discretized fourier space labelled by scalar wave vectors @xmath75 that are associated with the shells @xmath76 .
the dynamical variables are the complex scalar shell velocities @xmath77 , henceforth denoted by @xmath78 .
the evolution equations for the goy model are @xmath79u_n = \imath\bigg{[}a_n u_{n+1}u_{n+2 } + b_n u_{n-1}u_{n+1 } + c_n u_{n-1}u_{n-2 } \bigg{]}^{\ast } , \label{goy}\end{aligned}\ ] ] where @xmath80 , @xmath81 = 1/16 , complex conjugation is denoted by @xmath82 , and the coefficients @xmath83 , @xmath84 , @xmath85 are chosen to conserve the shell - model analogues of energy and helicity in the inviscid , unforced limit , and @xmath86 .
we use the standard value @xmath87 .
the second term allows for hyperviscous dissipation with degree @xmath41 ( the case @xmath40 corresponds to conventional viscous dissipation ) ; @xmath88 is a large wavenumber whose inverse is comparable to the dissipation length scale ; we choose @xmath89 and @xmath40 or @xmath90 .
since we concentrate on decaying turbulence here , the forcing term , required to drive the system into a statistically steady state , is absent .
the logarithmic discretization of fourier space allows us to reach very high reynolds numbers in numerical simulations of the goy model even with @xmath91 .
we use a slaved , second - order , adams - bashforth scheme @xcite to integrate the goy - model equations with @xmath92 shells , and , in some representative cases , @xmath93 ( this requires a fifth - order scheme @xcite ) , with the boundary conditions @xmath94 for @xmath95 and @xmath96 . in our simulations we use @xmath97 as the integration time step , and the viscosity @xmath98 for @xmath40 and @xmath99 for @xmath90 .
the goy - model equations allow for direct interactions only between nearest- and next - nearest - neighbour shells .
by contrast , in the fourier transform of the navier - stokes equation every fourier mode of the velocity is coupled to every other fourier mode directly .
this direct sweeping , as it is called , leads to the trivial dynamic scaling of eulerian velocity structure functions ( sect .
3 ) . since the goy model does not have direct sweeping in this sense , it is sometimes thought of as a highly simplified quasi - lagrangian version of the navier - stokes equation .
thus we might expect nontrivial dynamic multiscaling for goy - model structure functions .
the equal - time structure functions for this model are @xmath100^{p/2 } \ra \sim k_n^{-\zeta_p } , \label{goy1}\ ] ] where the power - law dependence is obtained only if @xmath101 lies in the inertial range .
three cycles @xcite in the static solutions of the goy model lead to rough , period - three oscillations in @xmath102 ; thus we use the modified structure function @xcite @xmath103|^{p/3}\ra \nonumber \\ & \sim & k_n^{-\zeta_p } , \label{sigma}\end{aligned}\ ] ] in which these oscillations are effectively filtered out
. therefore we use equations ( [ sigma ] ) and ( [ goy2 ] ) to extract @xmath28 .
time scales are obtained from the order-@xmath25 , time - dependent structure functions for the goy model , namely , @xmath104^{p/2 } \ra . \label{goy2}\ ] ] in refs .
@xcite we have used such time - dependent structure functions to verify the bridge relations ( [ zipm]-[zdpm ] ) for statistically steady turbulence . here
we give a short overview of our results for decaying turbulence with conventional viscosity @xmath105 and hyperviscosity @xmath106 . in our studies of decaying turbulence
we have used two types of initial conditions ; in both of these all the energy is initially concentrated in the first few fourier modes , i.e. , at large length scales , as in typical wind - tunnel experiments of homogeneous , isotropic turbulence @xcite . in the first initial condition @xmath107 , for @xmath76
= 1 , 2 , and @xmath108 for @xmath109 , where @xmath110 is a random phase angle distributed uniformly between 0 and @xmath111 ; we use this for the case @xmath40 .
the second initial condition we use is similar , namely , @xmath112 for @xmath86 .
we have used both these initial conditions for the case @xmath90 ; our results do not depend significantly on the initial condition that is used ; the results we report here for @xmath113 are for the second type of initial condition .
time is measured in terms of the initial large eddy - turnover time @xmath114 . for the goy shell model @xmath115 , where the root - mean - square velocity @xmath116^{1/2}$ ] is defined for the initial velocity field .
the shell - model energy spectrum is @xcite @xmath117 ; since @xmath118 shows the period - three oscillations mentioned above , a smooth energy spectrum is obtained by using @xmath119 as we show below .
the mean kinetic energy dissipation rate and the integral scale for the goy model are , respectively , @xmath120 and @xmath121 .
as the turbulence decays , @xmath122 increases ; once it becomes comparable to the size of the system the total energy decays @xcite as @xmath123 .
our results for dynamic - multiscaling exponents are obtained for times that are much shorter than the time over which @xmath122 becomes comparable to the system size ( @xmath124 for the goy model ) .
we begin by looking at the mean energy dissipation rate @xmath125 as a function of time .
a representative plot , averaged over 2000 initial conditions , is shown for @xmath90 in figure [ epsilon ] .
though this plot is noisy , it shows a clear peak .
this peak , at @xmath126 ( @xmath127 ) in figure [ epsilon ] , signals the completion of the cascade that transfers energy from the scale at which it is injected to the small scales where viscous dissipation becomes significant .
representative energy spectra at cascade completion are shown in figure [ spectrum ] ; after this point in time the energy spectrum decays very slowly without an _ appreciable _ change in the slope of the scaling regime , which appears as a nearly straight - line segment in the log - log plots of figure [ spectrum ] .
straight - line segments also show up in plots of the structure function @xmath128 at ( or after ) cascade completion as shown in the representative plots , for @xmath90 , of figure [ sigmaplot](a ) ; each curve in this figure has been averaged over 5000 independent initial conditions . from the slopes of the straight - line segments in these plots
( see table 1 ) we obtain the equal - time multiscaling exponents @xmath28 whose dependence on @xmath25 is shown in figure [ sigmaplot](b ) .
the values we quote for these ( and other ) exponents are the means of the slopes of 50 different plots like figure [ sigmaplot](a ) , which are obtained from 50 statistically independent runs ; the corresponding standard deviations yield the error bars shown in table 1 .
versus time for @xmath90 ( for clarity we show data for @xmath129 ) ; these data have been averaged over 2000 initial conditions .
the main peak at @xmath126 is a signature of cascade completion . for
@xmath40 @xmath125 displays a similar peak.,width=226,height=226 ] versus @xmath75 with the period - three oscillations ( see text ) .
the black dashed line corresponds to the k41 scaling prediction @xmath130 .
( b ) log - log plot of @xmath119 versus @xmath75 ; note that the period - three oscillations are suppressed here ; the black dashed line indicates the k41 scaling prediction.,title="fig:",width=226,height=226 ] versus @xmath75 with the period - three oscillations ( see text ) .
the black dashed line corresponds to the k41 scaling prediction @xmath130 .
( b ) log - log plot of @xmath119 versus @xmath75 ; note that the period - three oscillations are suppressed here ; the black dashed line indicates the k41 scaling prediction.,title="fig:",width=226,height=226 ] our results for the equal - time exponents @xmath28 for @xmath40 and @xmath90 are presented in table 1 . by comparing columns @xmath131 and @xmath132 in this table
we see that the exponents for both values of @xmath41 agree with each other and with the exponents reported earlier for statistically steady turbulence @xcite .
thus we reconfirm the universality of equal - time exponents : they neither depend on the precise dissipation mechanism nor on whether we consider statistically steady or decaying turbulence . , for @xmath133 to @xmath134 ( top to bottom ) , versus @xmath75 for @xmath90 and number of shells @xmath93 ( we show data only for the first @xmath135 shells ) .
( b ) plot of the equal - time multiscaling exponents @xmath28 ( obtained from ( a ) and listed in table 1 ) versus the order @xmath25 .
the open circles ( o ) , connected by a curve to guide the eye , indicate data from our numerical simulations ; the thick black line is the k41 prediction @xmath31.,title="fig:",width=226,height=226 ] , for @xmath133 to @xmath134 ( top to bottom ) , versus @xmath75 for @xmath90 and number of shells @xmath93 ( we show data only for the first @xmath135 shells ) .
( b ) plot of the equal - time multiscaling exponents @xmath28 ( obtained from ( a ) and listed in table 1 ) versus the order @xmath25 .
the open circles ( o ) , connected by a curve to guide the eye , indicate data from our numerical simulations ; the thick black line is the k41 prediction @xmath31.,title="fig:",width=226,height=226 ] lll order-@xmath25 & @xmath28 ( 4 - 14 ) & @xmath28 ( 4 - 16 ) + 1 & 0.380 @xmath136 0.001 & 0.37 @xmath136 0.01 + 2 & 0.709 @xmath136 0.003 & 0.699 @xmath136 0.008 + 3 & 1.000 @xmath136 0.005 & 1.003 @xmath136 0.008 + 4 & 1.266 @xmath136 0.008 & 1.29 @xmath136 0.02 + 5 & 1.51 @xmath136 0.01 & 1.55 @xmath136 0.03 + 6 & 1.74 @xmath136 0.02 & 1.79 @xmath136 0.05 + in ref .
@xcite the dynamic - multiscaling exponents @xmath137 and @xmath138 were obtained , by using integral- and derivative - time scales , for statistically steady turbulence in the goy model with @xmath139 .
we give below an overview of our recent results for dynamic multiscaling but for decaying turbulence ; for a detailed discussion of these results we refer the reader to ref .
@xcite .
time - dependent structure functions in decaying turbulence must , of course , depend on the origin of time @xmath140 from which we start making measurements .
it turns out that this dependence can be eliminated by normalising these structure functions by their values at the origin of time ; we have shown this analytically for the kraichnan model @xcite for passive - scalar turbulence and numerically in several other models @xcite .
we present some representative numerical results for the goy model below for which we use the normalised time - dependent structure functions @xmath141 it turns out that @xmath142 does not depend on @xmath140 as shown explicitly in figure [ qp ] for @xmath143 and @xmath144 .
this is why we have not displayed @xmath140 as an argument of @xmath142 . for @xmath145 obtained from 6 different values of @xmath140
; successive values of @xmath140 are separated by 0.5@xmath114 .
since the symbols used for different values of @xmath140 are indistinguishable on the scale of this figure , we conclude that @xmath146 does not depend on @xmath140.,width=226,height=226 ] given that @xmath142 does not depend on @xmath140 , we can use expressions such as ( [ zi]-[zd ] ) , with @xmath147 replaced by @xmath142 , to obtain dynamic - multiscaling exponents for decaying turbulence .
we consider here the order-1 integral- and order-2 derivative - time scales and the associated exponents @xmath148 and @xmath149 , respectively .
to calculate the integral - time scale @xmath150 , we evaluate the integral in equation ( [ timp ] ) with @xmath142 instead of @xmath147 and the upper limit replaced by @xmath151 , the time at which @xmath152 reaches a value @xmath153 $ ] .
we should , of course , use @xmath154 , i.e. , @xmath155 , but it is hard to obtain well - averaged data for large @xmath15 , so we use @xmath156 .
we have checked that our results are not affected significantly if we restrict ourselves to the range @xmath157 . in figure
[ q4](a ) we show representative plots of @xmath146 versus time @xmath158 for shells @xmath159 and @xmath160 and @xmath90 ; from this we obtain @xmath161 as described above .
the slopes of log - log plots of @xmath150 versus @xmath75 now yield @xmath148 as shown in figure [ q4](b ) for @xmath143 .
the derivative - time scale @xmath162 is calculated by using a sixth - order , centred difference scheme .
the derivative - time exponent @xmath149 is then obtained from slopes of log - log plots of @xmath162 versus @xmath75 .
tables 2 and 3 summarise our results for the dynamic - multiscaling exponents .
all these exponents have been calculated for @xmath75 in the inertial range @xmath163 ; the exponents @xmath149 are shown only for @xmath40 . in tables 2 and 3 we also give the exponents that we get by using the bridge relations ( [ zipm ] ) and ( [ zdpm ] ) along with the equal - time exponents given in table 1 ; there is good agreement between the exponents from our numerical simulations and those obtained via the bridge relations . moreover , the exponents @xmath148 and @xmath149 , for @xmath40 , and @xmath148 , for @xmath90 , which we obtain for decaying turbulence , are equal ( within error bars ) to their counterparts ( see table 2 of ref .
@xcite ) for statistically steady turbulence .
thus the dynamic - multiscaling exponents for the goy model are universal . plots of these exponents versus @xmath25 are shown in figures [ comp_integ](a ) and [ comp_integ](b ) . versus time @xmath158 for shells @xmath164 and @xmath160 ( from top to bottom ) .
( b ) a log - log plot of the integral - time scale @xmath161 versus @xmath75 ; the integral - time exponent @xmath165 follows from the slope of the line ( in the range @xmath166).,title="fig:",width=226,height=226 ] versus time @xmath158 for shells @xmath164 and @xmath160 ( from top to bottom ) .
( b ) a log - log plot of the integral - time scale @xmath161 versus @xmath75 ; the integral - time exponent @xmath165 follows from the slope of the line ( in the range @xmath166).,title="fig:",width=226,height=226 ] lll order-@xmath25 & @xmath148[eq.([zipm ] ) ] & @xmath148 + 1 & 0.630 @xmath136 0.001 & 0.62 @xmath136 0.01 + 2 & 0.675 @xmath136 0.009 & 0.683 @xmath136 0.003 + 3 & 0.69 @xmath136 0.01 & 0.712 @xmath136 0.004 + 4 & 0.71 @xmath136 0.02 & 0.728 @xmath136 0.006 + 5 & 0.74 @xmath136 0.05 & 0.750 @xmath136 0.009 + 6 & 0.76 @xmath136 0.08 & 0.76 @xmath136 0.01 + lllll order-@xmath25 & @xmath167[eq.([zipm ] ) ] & @xmath167 & @xmath168 $ ] & @xmath169 + 1 & 0.620 @xmath136 0.001 & 0.60 @xmath136 0.02 & 0.690 @xmath136 0.006 & 0.687 @xmath136 0.003 + 2 & 0.671 @xmath136 0.004 & 0.67 @xmath136 0.03 & 0.72 @xmath136 0.01 & 0.719 @xmath136 0.005 + 3 & 0.709 @xmath136 0.008 & 0.707 @xmath136 0.006 & 0.74
@xmath136 0.02 & 0.743 @xmath136 0.007 + 4 & 0.73 @xmath136 0.01 & 0.736 @xmath136 0.008 & 0.76 @xmath136 0.03 & 0.75 @xmath136 0.01 + 5 & 0.75 @xmath136 0.02 & 0.752 @xmath136 0.009 & 0.77 @xmath136 0.03 & 0.77 @xmath136 0.02 + 6 & 0.77 @xmath136 0.03 & 0.76 @xmath136 0.02 & 0.77 @xmath136 0.03 & 0.76 @xmath136 0.02 + the universality of dynamic exponents also goes through for models of passive - scalar turbulence as we show in refs .
it turns out that the kraichnan model @xcite shows simple scaling .
however , a shell - model version of a passive scalar advected by goy - model velocities exhibits dynamic multiscaling with dynamic exponents that depend on the degree @xmath54 , defined in equations ( [ timp]-[tdp ] ) , but not on the order @xmath25 .
we hope our detailed numerical simulations of dynamic multiscaling in the goy shell model for fluid turbulence will lead to experiments designed to measure time - dependent lagrangian or quasi - lagrangian structure functions in fluid turbulence .
advances in experimental techniques have made it possible to get good data for , say , the acceleration of lagrangian particles @xcite .
we believe such techniques can be refined to measure the sorts of time - dependent structure functions we have discussed here .
we would like to thank prasad perlekar and ganapati sahoo for discussions , ugc and dst ( india ) for support , and serc ( iisc ) for computational resources .
one of us ( rp ) is also part of the international collaboration for turbulence research ( ictr ) . and ( b ) @xmath149 versus @xmath25 ; the error bars are obtained as described in the text .
these plots compare data for statistically steady turbulence in the goy model ( from table 2 in ref .
@xcite ) and our data for decaying turbulence in the goy model with normal viscosity ( @xmath40 ) and hyperviscosity ( @xmath90 ) ; and they illustrate the universality of dynamic - multiscaling exponents that we discus in the text.,title="fig:",width=226,height=226 ] and ( b ) @xmath149 versus @xmath25 ; the error bars are obtained as described in the text .
these plots compare data for statistically steady turbulence in the goy model ( from table 2 in ref .
@xcite ) and our data for decaying turbulence in the goy model with normal viscosity ( @xmath40 ) and hyperviscosity ( @xmath90 ) ; and they illustrate the universality of dynamic - multiscaling exponents that we discus in the text.,title="fig:",width=226,height=226 ] 11 u. frisch , _ turbulence : the legacy of a.n . kolmogorov _
( cambridge university , cambridge , england , 1996 ) .
p.m. chaikin and t.c .
lubensky , _ principles of condensed matter physics _ ( cambridge university , cambridge , england , 2004 )
lvov , e. podivilov , and i. procaccia , phys .
e * 55*,7030 ( 1997 ) .
d. mitra and r. pandit , phys .
* 93 * , 2 ( 2004 ) .
d. mitra and r. pandit , phys .
lett . * 95 * , 144501 ( 2005 ) .
v.i . belinicher and v.s .
lvov , sov .
jetp * 66 * , 303(1987 ) .
y. kaneda , t. ishihara , and k. gotoh , phys .
fluids * 11 * , 2154 ( 1999 ) .
f. hayot and c. jayaprakash , phys .
e * 57 * , r4867 ( 1998 ) .
f. hayot and c. jayaprakash , int . j. mod .
b * 14 * , 1781 ( 2000 ) .
e. gledzer , sov .
. dokl . * 18 * , 216 ( 1973 ) .
k. ohkitani and m. yamada , prog .
* 81 * , 329 ( 1989 ) .
hohenberg and b.i .
halperin , rev .
phys . * 49 * , 435 ( 2004 ) and references therein .
kolmogorov , dokl .
nauk sssr * 30 * , 301 ( 1941 ) .
kolmogorov , dokl .
nauk sssr * 31 * , 538 ( 1941 ) .
e. leveque and z.s .
she , phys .
lett * 75 * , 2690 ( 1995 )
. n. schorghofer , l. kadanoff , and d. lohse , physica d * 88 * , 40 ( 1995 ) .
ditlevsen , phys .
fluids * 9 * , 1482 ( 1997 ) .
v. borue and s.a .
orzsag , phys .
e * 51 * , r856(1995 ) . v. s. lvov , i. procaccia , and d. vandembroucq , phys .
lett * 81 * , 802 ( 1998 ) .
lvov , r.a .
pasmanter , a. pomyalov , and i. procaccia , phys .
e * 67 * , 066310 ( 2003 ) .
pope , _ turbulent flows _ ( cambridge university , cambridge , england , 2000 ) . v. s. lvov and v. l. lebedev , phys .
e * 47 * , 1794 ( 1993 ) . s. s. ray , d. mitra , and r. pandit , to be published .
s. k. dhar , a. sain , and r. pandit , phys .
lett . * 78 * , 2964 ( 1997 ) .
d. pisarenko , l. biferale , d. courvoisier , u. frisch , and m. vergassola , phys .
fluids a * 5 * , 2533 ( 1993 ) .
g. sahoo , d. mitra , and r. pandit , to be published .
l. kadanoff , d. lohse , j. wang , and r. benzi , phys .
fluids * 7 * , 617 ( 1995 ) .
r. kraichnan , phys .
fluids * 11 * , 945 ( 1968 ) .
g. falkovich , k. gawedzki and m. vergassola , rev .
phys . * 73 * , 913 ( 2001 ) .
a. la porta , g.a .
voth , a.m. crawford , j. alexander , and e. bodenschatz nature * 409 * , 1017 ( 2001 ) . | we give an overview of the progress that has been made in recent years in understanding the dynamic multiscaling of homogeneous , isotropic turbulence and related problems .
we emphasise the similarity of this problem with the dynamic scaling of time - dependent correlation functions in the vicinity of a critical point in , e.g. , a spin system .
the universality of dynamic - multiscaling exponents in fluid turbulence is explored by detailed simulations of the goy shell model for fluid turbulence . |
an optical vortex is a light field exhibiting a pure screw phase dislocation along the propagation axis ; i.e. , an azimuthal phase dependence @xmath0 .
the integer @xmath1 plays the role of a topological charge : the phase changes its value by @xmath1 cycles of @xmath2 in any closed circuit about the axis , while the amplitude is zero there @xcite .
one of the most intriguing properties of vortices is that they carry orbital angular momentum ( oam ) .
this was first realized by allen and coworkers @xcite for the important instance of laguerre - gauss ( lg ) laser modes .
furthermore , they demonstrated that these modes carry an oam of @xmath3 per photon along the propagation direction .
a useful feature of optical oam is that it can be easily manipulated and transferred ; this has opened new horizons in various fields , ranging from mechanical micro - manipulation @xcite to imaging sciences @xcite , as well as potential astronomical @xcite and communication applications @xcite . beyond optical wavelengths ,
oam now plays a major role in electron @xcite , x - ray @xcite and radio frequency engineering @xcite .
the core observation that individual photons also carry oam brings the most exciting possibilities for employing this variable in the quantum regime , and a number of uses has already been demonstrated @xcite despite this intense activity , very little attention has been paid thus far to the radial index @xmath4 of the lg modes .
usually , it is stated that for @xmath5 , the modes are multiringed with @xmath6 radial nodes . beyond this short mention
, no physical meaning is attached to this quantity .
two recent papers , however , have presented challenging and interesting insights into this issue @xcite . our purpose here is to present a simple comprehensive analysis of this variable . the two aforementioned papers considered optical modes , governed by the paraxial wave equation .
these modes are ultimately suitably rescaled wave functions of the stationary states of a two - dimensional quantum oscillator , under the schrdinger equation @xcite . since
this latter system can properly model other interesting vortices arising in different media ( as in plasmas @xcite , superfluids @xcite , and bose - einstein condensates @xcite ) , the oscillator will serve as our thread , bearing in mind that the results can be immediately translated to the optical case .
to be as self - contained as possible , we briefly review the example of an isotropic two - dimensional quantum harmonic oscillator of mass @xmath7 and natural frequency @xmath8 , with coordinates in two orthogonal axes , say @xmath9 and @xmath10 @xcite .
the hamiltonian can be compactly written as @xmath11 , where the total number operator @xmath12 is @xmath13 and the annihilation and creation operators fulfil the canonical commutation relations @xmath14
= \delta_{jk } { \hat{\openone}}$ ] , with @xmath15 .
since the spectrum of @xmath16 is composed of all non - negative integers @xmath17 , the energies are given by @xmath18 and these eigenvalues are @xmath19-fold degenerate .
elements of the fock basis are the common eigenvectors of @xmath20 and @xmath21 : @xmath22 where @xmath23 is the ground state .
the stationary states of the oscillator are the product of hermite - gauss modes , as the oscillations in each axes are kinematically independent : @xmath24 \ , , \end{aligned}\ ] ] with @xmath25 .
to recover the equivalent beam solutions , one needs to take @xmath26 , since the paraxial wave equation ( in adimensional coordinates ) coincides with the schrdinger equation for the oscillator when @xmath27 and @xmath28 . for our purposes ,
the solution at @xmath29 is enough .
the wave function at any other time can be obtained in a simple way by using the explicit form of the propagator . for beams , where the role of time is played by the coordinate @xmath30 along the symmetry axis
, this propagation brings about additional interesting points , such as the gouy phase .
the axes @xmath9 and @xmath10 do not enjoy a privileged role in the problem .
since the energy is invariant under rotations in the @xmath31 plane , we could as well have chosen any other rotated reference frame . to take a better advantage of this symmetry , we consider the @xmath30-component of the angular momentum , @xmath32 , with @xmath33 , and use the rotated bosonic operators @xcite @xmath34 where @xmath35 = \delta_{jk } { \hat{\openone}}$ ] , with @xmath36 .
we can then check that @xmath37 whose interpretation is direct : the system can be envisioned now as consisting of `` quanta '' with positive ( counterclockwise rotation around @xmath30 ) and negative ( clockwise rotation around @xmath30 ) orbital angular momentum .
the fock basis @xmath38 of the common eigenvectors of @xmath39 and @xmath40 can be constructed much in the same way as in eq .
( [ eq:4 ] ) . however , it will prove useful to consider instead the continuous set @xmath41 parametrized by the complex number @xmath42 . the states @xmath43 constitute an orthonormal basis , whose properties have been reviewed in depth in ref .
@xcite . in the representation
they generate [ @xmath44 , the action of the basic operators is @xmath45 while for the adjoints we have @xmath46 since the exponential acting on the vacuum in eq .
is not unitary , the creation and destruction operators are not conjugates one of the other under the usual boson conjugation . given the above , @xmath12 and @xmath47 act in this space as @xmath48 as @xmath49 = 0 $ ] , the basis @xmath50 can be reinterpreted as common eigenvectors of @xmath12 and @xmath47 , with eigenvalues @xmath51 and @xmath52 , respectively .
the stationary states in this basis can be readily obtained using eqs .
( [ eq : nldif ] ) ; the final result is @xmath53 where the normalized amplitude is @xmath54 @xmath55 are the generalized laguerre polynomials and we have written @xmath56 .
the probability distribution @xmath57 shows @xmath4 dark concentric rings .
since the number of dark rings is @xmath58 , the operator @xmath59 seems to be a sensible definition for the radial - number operator of the laguerre - gauss modes . according to eq .
( [ eq : nldif ] ) , in differential form it reads @xmath60 incidentally , it coincides with the operator found in ref .
@xcite by setting @xmath61 . to simplify the following discussion ,
let us , for the time being , relabel the stationary states @xmath62 as @xmath63 , where @xmath4 indicates the radial mode eigenvalue ; i.e. , @xmath64 as heralded in the introduction , we are interested in exploring the hilbert space associated with the radial number @xmath4 , while keeping the oam @xmath1 fixed . at first sight , one might look for the canonical conjugate variable to @xmath65 . since , according to eq .
( [ eq : defp ] ) , @xmath66 or @xmath39 ( depending on the sign of @xmath1 ) , such a variable would be a phase @xmath67 or @xmath68 .
this means that if we denote by @xmath69 the exponential of such a putative phase , the corresponding commutation relation will read @xcite @xmath70 = { \hat{e } } \ , .\ ] ] this immediately implies @xmath71 so that @xmath72 whereas the spectrum of @xmath47 is unbounded , including all the integer numbers , the spectrum of @xmath65 is semibounded , as it comprises only non - negative integers .
this indicates that the action of @xmath73 as a ladder operator fails at @xmath74 , and consequently it can not be unitary : @xmath75 where @xmath76 is the projector on the `` vacuum . ''
all these problems thus place this interpretation on shaky grounds . for this reason
, we prefer to follow an alternative route . to this end
, we observe that to increase ( decrease ) the radial number by one unit , with @xmath1 unchanged , we need to create ( annihilate ) one positive quantum and one negative quantum , namely @xmath77 one can check that @xmath78 = - 2 { \hat{k}}_{z } \ , , \quad [ { \hat{k}}_{z } , { \hat{k}}_{+ } ] = { \hat{k}}_{+ } \ , , \quad [ { \hat{k}}_{z } , { \hat{k}}_{- } ] = - { \hat{k}}_{- } \ , , \ ] ] with @xmath79 .
this means that if we define @xmath80 , we have @xmath81 = - i { \hat{k}}_{z } \ , , \quad [ { \hat{k}}_{y } , { \hat{k}}_{z } ] = i { \hat{k}}_{x } \ , , \quad [ { \hat{k}}_{z } , { \hat{k}}_{x } ] = i { \hat{k}}_{y } \ , , \ ] ] that is , they are the generators of the su(1 , 1 ) algebra , as first noticed in ref .
@xcite . to explore the issue in more detail , it is convenient to give some basic background on some well - known irreducible representations ( irreps ) of su(1,1 ) , which are excellently reviewed in ref .
@xcite and whose role in quantum optics is difficult to underestimate @xcite .
the casimir operator for this group is @xmath82 , which can be expressed as @xmath83 , where the bargmann index @xmath84 labels the different irreps ( this index plays the role of spin for rotations ) . in our case
, a simple calculation shows that @xmath85 , so that @xmath86 , which corresponds to the so - called positive discrete series , for which @xmath87 is diagonal and has a discrete spectrum . in the fock basis @xmath38 , the basis states of the irrep @xmath84 are @xmath88 , with @xmath89 , and hence @xmath90 while the ladder operators act as @xmath91 note that we can make the identification @xmath92 , provided @xmath93 . since @xmath94
, this state can be taken as the vacuum .
indeed , @xmath95 are truly displacement operators , so according to perelomov prescription @xcite , the set @xmath96 constitutes a family of _ bona fide _ coherent states parametrized by the pseudo - euclidean unit vector @xmath97 , with @xmath98 and @xmath99 . by expanding the exponential and employing the disentangling theorem
, we get the decomposition @xmath100 and by projecting over the complete basis @xmath101 we get the corresponding wave function @xmath102 in the transverse parameters .
the expression can be simplified into an exponential form using the identity @xmath103 so the final result is @xmath104^{\frac{|\ell|+1}{2 } } \!\ !
e^{\frac{\zeta+1}{\zeta -1 } \,\alpha^{2 } r^2/2 } ( \alpha r)^{|\ell| } e^{i \ell\varphi } \ , .\ ] ] we see that the perelomov coherent states are polynomial - gauss modes at @xmath29 ; a subfamily of hypergeometric - gauss modes upon evolution , as discussed in ref .
they are also eigenstates of the oam , and shape invariant in the time evolution . for a coherent state @xmath105 with @xmath106 as as a function of @xmath4 and the ring average number @xmath107 . ]
the average number of sharp rings in the state @xmath108 is @xmath109 and the statistical distribution of rings @xmath110 is @xmath111 in fig . 1 we have plotted this distribution for a coherent state with @xmath106 and different values of @xmath107 . in fig .
[ fig : trans1 ] , we have plotted @xmath112 for a coherent state with @xmath106 and @xmath113 , and the corresponding distribution for the laguerre - gauss eigenstate with @xmath114 .
the striking differences can be appreciated at a glance .
it is worth mentioning that there is an alternative definition of coherent states , due to barut and girardello @xcite : @xmath115 which appears as a reasonable generalization of the standard coherent states as eigenstates of the annihilation operator .
this equation can be solved in the @xmath116 basis , yielding @xmath117 this set of coherent states are thus realized as bessel - gauss modes .
however , these solutions are not shape invariant , which runs against the notion of coherence . for a radial coherent state @xmath108 written as in , with @xmath118 and for an eigenstate @xmath119 with @xmath114 and @xmath106 . in the inset
, we show the corresponding density plots , in the same order . ] the commutation relations ( [ eq : ccr ] ) imply that these operators can not be measured simultaneously , which is reflected by the uncertainty relation @xmath120^{1/2}$ ] stands for the variance .
according to the standard definition , squeezing occurs whenever @xcite @xmath121 intelligent states are those for which ( [ eq : ur ] ) holds as an equality .
the coherent states ( [ eq : rcs ] ) and ( [ eq : bgproj ] ) are intelligent but not squeezed .
indeed , these intelligent states are solutions of the eigenvalue problem @xcite @xmath122 although they have been investigated from various perspectives @xcite , we follow here the comprehensive approach of ref
. @xcite , which starts by noting that the coherent state @xmath123 [ with eigenvalue @xmath124 , and @xmath125 an integer number ] .
then , the most general intelligent state can be written as @xmath126 where @xmath127 is the squeezing parameter and the seed state @xmath128 can be expressed as @xmath129 the coefficients @xmath130 can be obtained as a recursion relation ; the final result reads @xmath131 and @xmath132 is fixed by the normalization of the state .
the infinite family ( [ eq : puf ] ) of states parametrized by @xmath84 and @xmath133 is actually squeezed .
for intelligent states with @xmath134 and @xmath135 for several values of @xmath1 : @xmath136 , @xmath137 , and @xmath138 .
the inset shows the corresponding density plots , in the same order . ]
next , we need to project @xmath139 on the basis @xmath140 . to this end
, we recall that the action of @xmath141 on a basis state @xmath142 is given in terms of su(1,1 ) wigner @xmath143-functions @xcite @xmath144 where @xmath145 has been defined in eq . .
the effect of increasing @xmath1 , for fixed @xmath4 and @xmath127 , is illustrated by plotting the intensity profile @xmath146 as a function of @xmath147 in fig .
[ fig : variableell ] : this intensity tends to a gaussian - like shape .
the effect of increasing @xmath127 for fixed @xmath1 and @xmath133 is illustrated in fig .
[ fig : ffplot3d24 ] , and leads to the appearance of rings as we increase @xmath127 . for large values of @xmath1 ( more concretely , for @xmath84 and @xmath4 large , but @xmath148 )
, one has at hand a compact asymptotic approximation to the @xmath143 functions , namely @xcite @xmath149^{1/4 } } \,e^{-k(\tau-\tau_p)^2/2}\ , , \ ] ] with @xmath150 and whose gaussian nature is evident . for intelligent states with @xmath151 and @xmath152 for several values of @xmath127 :
@xmath153 , @xmath154 , and @xmath155 .
the inset shows the corresponding density plots , in the same order . ]
in summary , we have provided a handy toolbox to deal with the radial index of laguerre - gauss modes and shown how it can be used to construct a consistent quantum theory of this variable . we stress that this is more than an academic curiosity , since recent experiments in our laboratory @xcite have confirmed that the radial degree of freedom of single photons can be manipulated individually in a quantum regime .
it is our hope that the results presented here will inspire novel quantum protocols and algorithms using such a `` forgotten quantum number '' .
e. k. and r. w. b. acknowledge the support of the canada excellence research chairs ( cerc ) program .
the work of h. g. is supported by the natural sciences and engineering research council ( nserc ) of canada .
j. r. and z. h. are grateful to the financial assistance of the technology agency of the czech republic ( grant te01020229 ) and the czech ministry of industry and trade ( grant fr - ti1/364 ) .
g. l. is partially funded by eu fp7 ( grant q - essence ) .
finally , p. h. and l. l. s. s. acknowledge the support from the spanish mineco ( grant fis2011 - 26786 ) .
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 and , eds .
, @noop _ _ ( , , ) @noop * * , ( ) , , and , eds . , @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1038/nature08904 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) http://stacks.iop.org/1464-4258/6/i=5/a=020 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we introduce an operator linked with the radial index in the laguerre - gauss modes of a two - dimensional harmonic oscillator in cylindrical coordinates . we discuss ladder operators for this variable , and confirm that they obey the commutation relations of the su(1,1 ) algebra . using this fact
, we examine how basic quantum optical concepts can be recast in terms of radial modes . |
lattice qcd studies of hadron - hadron interactions are the gateway to nuclear physics through first principles @xcite . from a lattice simulation point of view
the nucleon - nucleon interaction is unquestionably the most challenging case @xcite , and it might not be resolved in the foreseeable future . however , interactions in other two - hadron systems are worth investigating as well , because new insights into the structural features of already discovered @xcite or yet unknown baryon resonances may emerge . in particular , one may ask if some of those may be understood as hadronic molecules , like the deuteron , or if more compact clusters , like a pentaquark @xcite , may also exist .
we are here interested in pairs of hadrons containing one heavy quark each . in such systems a relative , residual , interaction is a well - defined concept . in the spirit of the born - oppenheimer approximation ,
the ( slow ) heavy quarks serve to define the centers of the two hadrons while the ( fast ) light quarks and gluons provide the physics of the interaction .
exploratory studies along those lines have been done in the context of meson - meson and baryon - baryon systems @xcite .
specifically , we here investigate a heavy - light meson - baryon ( five - quark ) hadron with the quantum numbers of an s - wave @xmath0@xmath1 system .
the heavy - quark propagator is treated in the static approximation .
this allows us to compute the total energy as a function of the relative distance @xmath2 between the heavy quarks , viz .
a production scheme is illustrated in fig .
[ fig : klproduction ] .
the resulting adiabatic ( born - oppenheimer ) potential @xmath3 then may be used to address the possibility of molecule - like , or other , structures .
@xmath1 like molecule production and illustration of the current scheme for extracting a hadronic interaction at relative distance @xmath4 .
thick lines indicate heavy - quark propagators , and thin lines depict light quark propagators.,title="fig:",width=207 ] + @xmath1 like molecule production and illustration of the current scheme for extracting a hadronic interaction at relative distance @xmath4 .
thick lines indicate heavy - quark propagators , and thin lines depict light quark propagators.,title="fig:",width=207 ] in the static approximation to the heavy - quark propagator the potential @xmath3 extracted from the lattice equally applies to the systems @xmath0@xmath1 , @xmath5@xmath6 , and @xmath7@xmath8 , the @xmath9-flavored quarks being replaced with @xmath10 and @xmath11 flavors , respectively . within the framework of a schrdinger equation those systems
can be studied using their respective physical reduced masses , provided of course , one is willing to accept the limitations of the adiabatic approximation and the non - relativistic nature of the framework .
this is the main reason for classifying our results as qualitative .
the lattice simulation requires two - hadron interpolating fields . those are constructed using as building blocks standard local operators for the @xmath12 and @xmath13 particles @xcite .
they are placed at relative distance @xmath4 and then projected to total momentum zero @xmath14 here the normalization @xmath15 is the spatial lattice volume , the sums are over lattice sites , and @xmath16 is a dirac spinor index .
then , with @xmath17 the time correlation function @xmath18 where @xmath19 is the relative distance at the source , can be expressed in terms of fermion propagators .
the flavor assignment @xmath20 causes the separable term in ( [ corfun ] ) to vanish .
writing @xmath21 and @xmath22 for the heavy ( s ) and light ( u , d ) quark propagators , respectively , one obtains @xmath23\times\nonumber\\ & & \phantom{-}g(\vec{y}t,\vec{r}_1t_0 ) [ g(\vec{y}t,\vec{r}_1t_0)g(\vec{y}+\vec{r}t,\vec{r}_1+\vec{s}t_0)\nonumber\\ & & -g(\vec{y}t,\vec{r}_1+\vec{s}t_0)g(\vec{y}+\vec{r}t,\vec{r}_1t_0)]\;\rangle\,.\end{aligned}\ ] ] for clarity the rather involved color and spin index structure is not shown in ( [ chg ] ) .
also , translational invariance of the gauge field average @xmath24 has been used to arrive at the above expression , and utilizing this freedom , an arbitrary space site @xmath25 was introduced to fix the source locations . a diagrammatic representation of ( [ chg ] ) is shown in fig . [ fig : chg ] . ' '' '' ) .
thick lines indicate heavy - quark propagators , and thin lines depict light ( u and d ) quark propagators.,title="fig:",width=102 ] ) .
thick lines indicate heavy - quark propagators , and thin lines depict light ( u and d ) quark propagators.,title="fig:",width=102 ] + ) .
thick lines indicate heavy - quark propagators , and thin lines depict light ( u and d ) quark propagators.,title="fig:",width=102 ] ) .
thick lines indicate heavy - quark propagators , and thin lines depict light ( u and d ) quark propagators.,title="fig:",width=102 ] the heavy - quark propagators are employed in the static approximation . for ( unimproved )
wilson fermions with hopping parameter @xmath26 this means that the propagator is taken in the limit @xmath27 , resulting in @xmath28 where @xmath29 is the product of @xmath30 link variables along a straight line from @xmath31 to @xmath32 @xcite .
the factor @xmath33 gives rise to only a constant global energy shift @xmath34 , which we will ignore .
the distance @xmath35 is rather special @xcite because a color singlet operator , as realized by ( [ op2 ] ) , can also be achieved by a `` color twisted '' version of ( [ op2 ] ) where quarks across the hadrons @xmath12 and @xmath13 , now at the same location , are combined into a color singlet . because we do not consider color twisted operators in this work , we restrict ourselves to non - zero relative distance . ] .
thus , using @xmath36 and @xmath37 , the first two terms in ( [ chg ] ) , and accordingly the top two diagrams in fig .
[ fig : chg ] vanish , and only the last two of those make a contribution to the correlation function for non - zero relative distance . by way of ( [ hstatic ] )
those contributions are proportional to @xmath38 . introducing @xmath39 , to replace @xmath19 ,
the correlation function ( [ chg ] ) becomes @xmath40\;\rangle \quad\mbox{for}\quad\vec{r}\neq 0\nonumber\,.\end{aligned}\ ] ] as a consequence of the static approximation , the site sum @xmath41 has vanished from ( [ chg ] ) , and thus , is unfortunately no longer working to improve statistics .
the final correlator we use is extended from ( [ c ] ) to a @xmath42 matrix by employing several levels @xmath43 of operator smearing .
the procedure amounts to replacing in ( [ op2 ] ) all light - quark fields @xmath44 with smeared fields @xmath45 .
we have used wuppertal - style fermion smearing @xcite along with ape - style gauge field fuzzing @xcite .
the implementation specific to the asymmetric lattice used here is discussed below .
no smearing , nor link variable fuzzing , was done for the heavy , static , quark fields in order to preserve spatial locality , i.e. the @xmath46 factor in ( [ hstatic ] ) .
thus , generically , replacing @xmath47 $ ] the correlator ( [ c ] ) becomes a @xmath42 matrix @xmath48 with @xmath49 and a sum over @xmath50 is understood .
the expression for @xmath51 in terms of quark propagators still has the form given by ( [ c ] ) , however , light - quark propagator elements are replaced with smeared ones , @xmath52 . smearing and fuzzing prescriptions are the same at source and sink . thus the correlation matrix ( [ ckk ] ) is hermitian by construction .
we choose an asymmetric and anisotropic lattice with geometry @xmath53 .
the bare lattice constants , in the respective directions , satisfy @xmath54 .
subsequently we shall refer to @xmath55 as the lattice constant @xmath56 of the simulation , so the physical volume is @xmath57 .
the lattice has a fine resolution in 4-direction ( time ) , and the same fine resolution in the ( spatial ) 3-direction where the lattice size is twice as long as in the 1,2-directions .
the idea here is to place the static quark sources along the 3-direction and thus achieve a fine spatial resolution of the relative distance as well as to provide enough space to allow for asymptotic separation of the hadrons .
the spatial lattice asymmetry leads us to modify the procedure for operator smearing .
a smearing iteration of @xmath58 steps is understood to mean @xmath58 steps in the 1,2-directions but @xmath59 steps in the 3-direction , so the physical smearing extent is the same for all spatial directions ( neglecting renormalization effects ) .
mutatis mutandis , the same is true for gauge field fuzzing .
an illustration is shown in fig .
[ fig : asym ] . the implementation used here
is described in more technical terms in reference @xcite .
iteration is shown.,title="fig:",width=170 ] + iteration is shown.,title="fig:",width=170 ] + the positions of the static quark sources are chosen as @xmath60 with @xmath61 .
their spatial locations along the 3-direction , see fig .
[ fig : rdistcirc ] , allow us to achieve any relative distance @xmath62 by choosing an appropriate pair of source points .
some relative distances can be obtained more than once . in those cases
appropriate averages over sets of source and sink pairs have been taken to compute the correlator matrix elements . because periodic boundary conditions are applied in the spatial directions relative distances larger than @xmath63 are redundant . and
@xmath63 can be achieved , some multiple times.,width=321 ] given the approximations employed in this work , e.g. static heavy quarks , and a few more mentioned below , it would be of no advantage to employ sophisticated improved lattice actions .
we have therefore used the standard wilson plaquette action with wilson fermions in a quenched simulation @xcite .
the gauge field couplings in the @xmath50-@xmath64 planes and the hopping parameters in directions @xmath50 are given by , respectively , @xmath65 the simulation was done at @xmath66 with four values @xmath67 of the hopping parameter employing a multiple mass solver @xcite .
a total of 343 gauge field configurations were used .
the time evolution of the eigenvalues of @xmath68 determines the energy spectrum .
eigenvalues , behaving exponentially with @xmath69 , may rapidly vanish into numerical noise .
conventional diagonalization methods do not work well under those circumstances .
singular value decomposition ( svd ) , on the other hand , is ideally suited to the task @xcite .
the svd reads @xmath70 where @xmath71 and @xmath72 are unitary in our case , and @xmath73 contains the singular values satisfying @xmath74 .
if @xmath68 is non - degenerate and positive semi - definite then the set of singular values @xmath75 and the set of eigenvalues are the same .
a few more details can be found in @xcite . for simplicity we will refer to @xmath76 as eigenvalues .
to extract energy levels from the eigenvalues an often used procedure , sometimes starting from a generalized eigenvalue problem @xcite , is to construct effective mass functions from the eigenvalues and then try to identify plateaus in the asymptotic time region .
typically , only a narrow subset of the time slices is usable , which also is subject to some discretion .
we shall not rely on effective mass functions here , but rather analyze correlator eigenvalues by means of the maximum entropy method @xcite following the adaption described in @xcite , and in @xcite . among the advantages
are that all available time slices may be used , if desired , and that excited states , if present , are easily revealed .
the time dependence of the eigenvalues @xmath77 shall be fit with the model @xmath78 for some set of discrete times slices @xmath79 , where @xmath80 is a spectral density function .
the maximum entropy method ( mem ) is based upon bayesian statistics . in this context
the numbers @xmath80 are interpreted as stochastic variables @xcite .
their most likely values are obtained by simulated annealing as described in @xcite .
we refer the reader to this reference for technical details . at each of the @xmath81 relative distances @xmath2
we have used three ( @xmath82 ) operator smearing levels @xmath83 . a typical set of eigenvalues
is displayed in fig .
[ fig : corval ] for distance @xmath84 and the smallest ( light ) quark mass , @xmath85 .
only two eigenvalue correlators are shown , the third one is indistinguishable from zero due to numerical noise . note that eigenvalues are separated by about four orders of magnitude . thus crossing ambiguities are nonexistent .
the eigenvalue sets for all other distances @xmath2 exhibit the same features . of the @xmath86 time correlation matrix ( [ svd ] ) for relative distance @xmath84 and quark hopping parameter @xmath85 .
the third eigenvalue , @xmath87 , is not shown , it vanishes into numerical noise .
filled plot symbols belong to time slices used for the mem fits employing the model ( [ fmodel ] ) , the curves are the corresponding results .
note that the source has been shifted to @xmath88.,width=241 ] ' '' ''
the eigenvalue correlators were fit with the model ( [ fmodel ] ) using , consistently in all cases , time slices @xmath89 with the source @xmath88 being omitted , and mass range of @xmath90 with discretization interval size @xmath91 , all in units of @xmath92 .
the maximum entropy method then yields spectral density functions @xmath80 .
representative examples are shown in fig .
[ fig : memspec ] for @xmath93 , in units of @xmath56 , and @xmath85 .
the histogram lines in fig .
[ fig : memspec ] represent @xmath80 , obtained from simulating annealing ( cooling ) following the very procedure put forward in @xcite .
sixteen random annealing starts were used .
the solid histogram lines represent the average of those 16 runs , the dashed histogram lines give the corresponding standard deviation . , according to ( [ fmodel ] ) , for two - hadron operators with relative distance @xmath93 , in units of @xmath56 , thick histogram lines . shown
are eigenvalue spectra which contain ground state peaks .
the dashed histogram lines indicate errors as explained in the text .
the smooth dotted lines correspond to gaussian fits to the discretized spectral densities.,title="fig:",width=257 ] + , according to ( [ fmodel ] ) , for two - hadron operators with relative distance @xmath93 , in units of @xmath56 , thick histogram lines . shown
are eigenvalue spectra which contain ground state peaks .
the dashed histogram lines indicate errors as explained in the text .
the smooth dotted lines correspond to gaussian fits to the discretized spectral densities.,title="fig:",width=264 ] + , according to ( [ fmodel ] ) , for two - hadron operators with relative distance @xmath93 , in units of @xmath56 , thick histogram lines .
shown are eigenvalue spectra which contain ground state peaks .
the dashed histogram lines indicate errors as explained in the text .
the smooth dotted lines correspond to gaussian fits to the discretized spectral densities.,title="fig:",width=257 ] + , according to ( [ fmodel ] ) , for two - hadron operators with relative distance @xmath93 , in units of @xmath56 , thick histogram lines . shown
are eigenvalue spectra which contain ground state peaks .
the dashed histogram lines indicate errors as explained in the text .
the smooth dotted lines correspond to gaussian fits to the discretized spectral densities.,title="fig:",width=260 ] in most cases the spectra exhibit isolated peaks , say @xmath94 .
then one may compute , for each peak @xmath95 , the volume @xmath96 , the energy @xmath97 , and the width @xmath98 , according to @xmath99 typically the largest eigenvalues reveal the ground state .
there are few exceptions .
the spectra at @xmath84 and @xmath100 , shown in fig .
[ fig : memspec ] , are examples . selecting the support @xmath101 of a peak can be slightly ambiguous , particularly if the peak sits on background noise , or if there is overlap with another one .
we have therefore , as a matter of course , performed least - square fits to the average spectral functions @xmath80 with one or two gaussians , as required in each case .
these fits then give values for @xmath96 , @xmath97 , and @xmath98 as defined in ( [ zn])([dn ] ) directly from the parameters of the gaussians . in this manner
we have obtained 16 sets , one for each @xmath2 , of five - quark hadron ground state energies .
each set contains four energies corresponding to hopping parameter values @xmath67 .
the spectral density functions @xmath80 typically exhibit some additional structure at the high end of the @xmath102 range , a broad peak in most cases .
these secondary peaks appear because diagonalizing the correlator matrix separately on all time slices only ensures that its eigenvalues describe a single state from the physical spectrum at large values for @xmath69 @xcite . at small time slices , close to the source
, the eigenvectors of @xmath68 may still describe a mix of spectral states .
in practice the large-@xmath69 rule means that , in each eigen channel , one should only take the lowest peak into account . in physical terms
the secondary peaks are separated from those by @xmath1033gev , and larger , and thus should probably be considered lattice artifacts . in order to relate our results to the physical hadron spectrum we have also computed @xmath86 correlation matrices with standard local @xcite
pseudoscalar meson ( @xmath104 ) , vector meson ( @xmath105 ) , and nucleon ( @xmath15 ) operators , using the same smearing and fuzzing prescription . the analysis was performed with the mem in exactly the same way as described above .
this now allows extrapolations of hadron masses to the physical pion mass region ( @xmath106 ) .
we use the extrapolation model introduced in @xcite @xmath107 and @xmath108 . for a motivation of the same we refer the reader to said reference .
the extrapolation prescription is , of course , a source of systematic error on our results . in fig .
[ fig : extra ] we show the vector meson and nucleon masses as a function of the squared pion mass and the extrapolations according to ( [ mpi0 ] ) .
the extrapolated values @xmath109 and @xmath110 , as @xmath111 , are used to set the reduced mass of the @xmath112 system to its experimental value @xmath113mev .
this results in a lattice constant of @xmath114mev or @xmath115fm .
the corresponding vector meson and nucleon masses come out as 659mev and 1196mev , respectively , which deviate by @xmath116 and @xmath117 from their experimental values .
we take these numbers as indicators of the systematic errors to be expected due to the extrapolation . ' '' '' based on the model ( [ mpi0 ] ) and four hopping parameter values , top panel .
the bottom panel shows two typical examples of five - quark hadron ground state energies @xmath118 , at distances @xmath84 and @xmath119 , and their respective extrapolation using the same procedure.,title="fig:",width=287 ] + based on the model ( [ mpi0 ] ) and four hopping parameter values , top panel .
the bottom panel shows two typical examples of five - quark hadron ground state energies
@xmath118 , at distances @xmath84 and @xmath119 , and their respective extrapolation using the same procedure.,title="fig:",width=291 ] also in fig .
[ fig : extra ] two representative examples of extrapolations of five - quark hadron ground state masses @xmath118 are shown , the relative distances are @xmath84 and @xmath119 .
the error bars represent spectral peak widths @xmath98 according to ( [ dn ] ) .
we utilize those to compute uncertainties on the extrapolated energies : at each of the four data points stochastically independent gaussian random numbers with average and standard deviation given by the energies and peak widths of the data , respectively , were generated and again fitted with ( [ mpi0 ] ) . repeating this a large number of times then gives rise to a standard deviation which we interpret as a mem peak extrapolated width @xmath98 . the extrapolated energy spectrum obtained in this way
is displayed in fig .
[ fig : wpot ] as a function of the relative distance @xmath2 . of the heavy - light @xmath0@xmath1 like system versus the relative distance @xmath2 with mem based uncertainties ( plot symbols ) .
the solid line represents a fit with the model ( [ vl ] ) .
the dashed line shows the same fit , but the mirror term @xmath120 has been dropped.,width=321 ] the energies shown in fig .
[ fig : wpot ] are also subject to statistical errors which originate from the gauge configuration ensemble . to identify those a bootstrap error analysis @xcite
was performed . using a resampling factor of eight on the 343 gauge configurations
the mem spectral analysis was repeated for each bootstrap sample . in each case
the same fixed random number sequence was employed in order to eliminate fluctuations in the annealing process .
the statistical errors @xmath121 thus obtained are compared to the peak widths @xmath98 in tab .
[ tab : serr ] .
results are for @xmath85 ( lightest quark mass ) .
the other @xmath26 values give very similar numbers because the underlying set of gauge configurations is the same .
clearly the statistical errors @xmath121 are consistently smaller than the peak widths @xmath98 . on average over all relative distances @xmath2
the ratio @xmath122 turns out to be slightly larger than two .
thus , judiciously keeping the larger errors , we have used the spectral peak widths as the principle input for computing uncertainties .
rccc @xmath123 & @xmath124 & @xmath125 & @xmath126 + 1 & 1.193 & 0.050 & 0.038 + 2 & 1.434 & 0.106 & 0.031 + 3 & 1.514 & 0.083 & 0.037 + 4 & 1.526 & 0.062 & 0.059 + 5 & 1.512 & 0.067 & 0.076 + 6 & 1.378 & 0.053 & 0.032 + 7 & 1.206 & 0.052 & 0.024 + 8 & 1.120 & 0.035 & 0.029 + 9 & 1.190 & 0.055 & 0.030 + 10 & 1.223 & 0.065 & 0.020 + 11 & 1.348 & 0.066 & 0.034 + 12 & 1.390 & 0.072 & 0.030 + 13 & 1.465 & 0.047 & 0.024 + 14 & 1.338 & 0.080 & 0.030 + 15 & 1.315 & 0.058 & 0.040 + 16 & 1.297 & 0.062 & 0.009
the data points in fig . [ fig : wpot ] connect smoothly along a continuous curve . in order to find an appropriate fit model we note that the ground - state @xmath0@xmath1 like system should be in a relative s - wave .
our choice thus are s - wave 3d harmonic oscillator functions @xcite .
the first few terms are given by @xmath127 with parameters @xmath128 .
this will not be sufficient though for two reasons .
first , periodic boundary conditions have been imposed on the lattice in the 3-direction . thus lattice sites with relative distances @xmath129 and @xmath130 are equivalent . however , as the two - hadron operators for the @xmath0@xmath1 like system are essentially loop operators , see fig . [
fig : chg ] , those do not share the periodic behavior of the fundamental lattice fields .
this effect should be most notable for distances around @xmath131 .
second , interactions with hadron images in adjacent copies of the lattice are present .
their dominant effect is a constant interaction energy , as the relative distances of most of the mirror images remain constant with varying @xmath129 .
thus motivated , empirically one finds that the lattice data are well represented by the model @xmath132 where @xmath133 and @xmath134 are additional parameters .
the resulting fit is shown in fig .
[ fig : wpot ] as a solid line .
the dashed line represents @xmath135 only . table [ tab : wparam ] contains the fit parameters .
cd @xmath133 & 1.213(93 ) + @xmath136 & 0.0409(37 ) + @xmath137 & -0.89(10 ) + @xmath138 & 0.109(19 ) + @xmath139 & -0.00328(52 ) + @xmath134 & 26.54(57 ) note that @xmath134 deviates somewhat from @xmath140 , as should be expected for the reason stated above .
the @xmath141 per degree of freedom is @xmath142 , the uncertainties given stem from the covariance matrix of the levenberg - marquardt algorithm .
we list those for completeness only , they will not be used in the subsequent analysis .
rather , because we are mostly interested in uncertainties of possible bound state energies , as the case may be , we proceed as follows : each data point in fig .
[ fig : wpot ] is randomized by stochastically independent gaussian random numbers with average and standard deviation given by the energies and peak widths of the data , respectively .
then the fit with the model ( [ vl ] ) is repeated , the corresponding set of parameters gives rise to a new potential @xmath143 .
the procedure is performed 32 times .
we show the results in fig .
[ fig : vpot ] along with the potential obtained directly from the original data points .
specifically , fig .
[ fig : vpot ] displays adiabatic potentials @xmath144 on physical scales given by the lattice constant @xmath56 , as determined above .
resulting from the lattice simulation in physical units , panel ( a ) .
panel ( b ) shows potentials from 32 sets of randomizations of the ground state energies @xmath97 as seen in fig .
[ fig : wpot].,title="fig:",width=321 ] + resulting from the lattice simulation in physical units , panel ( a ) . panel ( b ) shows potentials from 32 sets of randomizations of the ground state energies @xmath97 as seen in fig .
[ fig : wpot].,title="fig:",width=321 ]
figure [ fig : vpot ] represents the end result of the lattice simulation . although we expect that systematic errors coming from extrapolations may be as large as @xmath145 , and the uncertainties on the potential itself according to fig .
[ fig : vpot](b ) are sizable , we believe that important physics can still be learned on a qualitative level .
first , we observe that @xmath3 of fig . [
fig : vpot](a ) has no repulsive regions .
this is a reasonable , and confidence inspiring , outcome because repulsion in a two - hadron system is typically pauli repulsion , but of course the latter is absent in a @xmath0@xmath1 like system .
also , as mentioned earlier just before eqn .
( [ c ] ) , the relative distance @xmath146 was excluded from the simulation because `` color twisted '' operators would then have to be included for all distances @xmath2 , to be consistent . without further work it remains an open question how such a family of operators might influence the form of @xmath3 .
however , it seems reasonable to assume that those will not play a major role for large @xmath2 because there they would create hadrons inconsistent with confinement , whereas at @xmath146 mixing with those states would lower the ground state energy according to general principles of quantum mechanics .
thus , we believe that the current simulation correctly captures the most important physics of the four - quark system .
the salient features of @xmath3 are the two distinct regions of attraction around @xmath147@xmath148fm and @xmath149fm .
an obvious question is whether the attraction is sufficient to generate bound states of the five - quark molecule . to find out we have , first , computed scattering phase shifts @xmath150 employing the adiabatic potential @xmath3 in a schrdinger equation . for the reduced mass @xmath151 we chose , in turn , the three values of the physical @xmath0@xmath1 , @xmath5@xmath6 , and @xmath7@xmath8 systems @xcite .
these evaluate to @xmath152 , and @xmath153 , respectively .
continuum boundary conditions were implemented by solving a volterra integral equation within the jost function formalism @xcite .
the resulting scattering phase shifts are shown in fig .
[ fig : scatt ] . according to levinson s theorem ,
see @xcite for example , we have @xmath154 where @xmath95 is the number of bound states . in the s - wave @xmath95
is one for the @xmath0@xmath1 system , two for the @xmath5@xmath6 system , and four for the @xmath7@xmath8 system .
we need to caution though that relativistic effect are large in @xmath0@xmath1 , with increasing quark mass ( @xmath155 ) this becomes less of a concern .
the scattering phase shifts are quite featureless , but they give us an exact count of the number of bound states present in the simulation .
@xmath1 , @xmath5@xmath6 , and @xmath7@xmath8 systems ( bottom to top , respectively ) versus the relative momentum @xmath156.,width=309 ] in order to compute bound state energies and wave functions we solve the radial schrdinger equation with @xmath3 .
our conventions can be gleaned from @xmath157u_a=2meu_a\,,\ ] ] and @xmath158 the region between @xmath146 and @xmath159 was subdivided into 60 discretization intervals .
a suitable cubic b - spline representation @xcite for @xmath160 was employed to enforce bound - state boundary conditions and provide smooth interpolation between the knots .
this then leads to a generalized matrix eigenvalue problem . the scale neutral bound state energies are compiled in tab .
[ tab : bstate ] for the three reduced masses of @xmath0@xmath1 , @xmath5@xmath6 , @xmath7@xmath8 , and three partial waves .
the errors come from repeating the calculation with 32 randomizations of the potential , as seen in fig .
[ fig : vpot](b ) .
cccccc @xmath161 & @xmath162 & @xmath163 & @xmath164 & @xmath165 & @xmath166 + 0 & 0.152 & -0.22(12 ) & & & + & 0.415 & -0.32(13 ) & -0.04(13 ) & & + & 1.172 & -0.39(13 ) & -0.26(16 ) & -0.15(12 ) & -0.001(88 ) + 1 & 0.152 & -0.11(11 ) & & & + & 0.415 & -0.28(13 ) & & & + & 1.172 & -0.38(13 ) & -0.14(12 ) & & + 2 & 0.152 & & & & + & 0.415 & -0.20(13 ) & & & + & 1.172 & -0.35(13 ) & -0.11(12 ) & & the s - wave bound state wave functions @xmath160 and their squares @xmath167 are displayed in figs .
[ fig:5s],[fig:5c ] and [ fig:5b ] .
the wave function of the @xmath0@xmath1 system , fig .
[ fig:5s ] , peaks at around @xmath168@xmath169fm .
this is a typical nuclear physics distance scale , and clearly points to the structure of a hadronic molecule
. with increasing ( heavy ) quark mass the number of bound states also increases .
the ground state of the @xmath5@xmath6 system , fig .
[ fig:5c ] , exhibits the same feature .
however , the excited state wave function has sizable support at small distances @xmath149fm .
the wave function starts to ` feel ' the short - range attraction of the potential @xmath3 , see fig .
[ fig : vpot](a ) .
this property is fully developed in @xmath7@xmath8 , see fig .
[ fig:5b ] .
the ground state wave function ` lives ' in the large - distance trough of @xmath3 and thus points to a hadronic molecule . on the other hand , the wave function of first excited state ` lives ' in the small - distance regime of @xmath3 , which indicates a very tight spacial structure .
there is nothing as distinct about the third state , and the forth one strongly resembles a continuum state .
( dashed line ) and its square @xmath170 ( solid line ) for the @xmath0@xmath1 system.,width=272 ] ( dashed lines ) and their squares @xmath170 ( solid lines ) for the @xmath5@xmath6 system.,title="fig:",width=272 ] + ( dashed lines ) and their squares @xmath170 ( solid lines ) for the @xmath5@xmath6 system.,title="fig:",width=272 ] ( dashed lines ) and their squares @xmath170 ( solid lines ) for the @xmath7@xmath8 system.,title="fig:",width=272 ] + ( dashed lines ) and their squares @xmath170 ( solid lines ) for the @xmath7@xmath8 system.,title="fig:",width=272 ] + ( dashed lines ) and their squares @xmath170 ( solid lines ) for the @xmath7@xmath8 system.,title="fig:",width=272 ] + ( dashed lines ) and their squares @xmath170 ( solid lines ) for the @xmath7@xmath8 system.,title="fig:",width=272 ] the wave functions of the ground and first excited @xmath7@xmath8 states with @xmath171 and @xmath172 , respectively , shown in fig .
[ fig:5b ] deserve further comment .
viewing the system as consisting of five quarks , the ground state is clearly a pure two - hadron molecule , and the rms - radius evaluates to @xmath173fm .
the excited state , on the other hand , is spatially compact in a dramatic way : most of the probability density @xmath167 is within a region of less than @xmath174fm radius and peaks just above @xmath175fm .
inevitably , this five - quark state should be interpreted as a pentaquark .
we should remark here that our original five - quark operators ( [ op2 ] ) are indeed capable of describing such physics , because the set comprises operators where the heavy quark and anti - quark have very small distances , the smallest one being the lattice constant @xmath115fm . also , the @xmath7@xmath8 results are probably our most reliable ones , because relativistic effects are not very important . on the other hand , systematic errors , having to do with the adiabatic approximation , extrapolations to zero pion mass , and dismissal of relativistic effects , clearly render our interpretation a qualitative one .
it is interesting to ask though if the reason for our inability to establish a clear experimental signal for a pentaquark @xcite , and also to irrefutably establish a lattice qcd pentaquark @xcite , could be rooted in the assumption that the pentaquark shall be a ground state of bound light quarks .
there is no a priori reason for this .
the well - publicized quark model decouplet @xmath176 particle @xcite may well have a molecule - like structure .
thus we speculate that the pentaquark , if we understand it as a tight five - quark bound state , may not exist for all light flavored ( u , d , s ) quarks , but there is still hope in the heavy - flavor sector .
however , a pentaquark may then reveal itself only as an excited state , the ground state being a hadronic molecule . finally , as a matter of course ,
we show in figs .
[ fig:6p ] and [ fig:6d ] the @xmath177- and @xmath5-state wave functions of the calculation . and in tab .
[ tab : bstatemasses ] we have compiled physical mass spectra of the considered five - quark systems by subtracting the computed binding energies from the experimental single meson and baryon masses @xcite .
we give those numbers for completeness only , without claiming quantitative relevance .
( dashed lines ) and their squares @xmath170 ( solid lines ) .
the insets for @xmath162 and @xmath178 match the entries of tab .
[ tab : bstate].,title="fig:",width=272 ] + ( dashed lines ) and their squares @xmath170 ( solid lines ) . the insets for @xmath162 and @xmath178 match the entries of tab .
[ tab : bstate].,title="fig:",width=272 ] + ( dashed lines ) and their squares @xmath170 ( solid lines ) . the insets for @xmath162 and @xmath178 match the entries of tab .
[ tab : bstate].,title="fig:",width=272 ] + ( dashed lines ) and their squares @xmath170 ( solid lines ) .
the insets for @xmath162 and @xmath178 match the entries of tab .
[ tab : bstate].,title="fig:",width=272 ] ( dashed lines ) and their squares @xmath170 ( solid lines ) .
the insets for @xmath162 and @xmath178 match the entries of tab .
[ tab : bstate].,title="fig:",width=272 ] + ( dashed lines ) and their squares @xmath170 ( solid lines ) .
the insets for @xmath162 and @xmath178 match the entries of tab .
[ tab : bstate].,title="fig:",width=272 ] + ( dashed lines ) and their squares @xmath170 ( solid lines ) .
the insets for @xmath162 and @xmath178 match the entries of tab .
[ tab : bstate].,title="fig:",width=272 ] cccccc @xmath161 & @xmath151 & @xmath179 & @xmath180 & @xmath181 & @xmath182 + 0 & 0.342 & 1.11(28 ) & & & + & 0.934 & 3.44(29 ) & 4.07(29 ) & & + & 2.639 & 10.02(30 ) & 10.33(35 ) & 10.58(27 ) & 10.90(20 ) + 1 & 0.342 & 1.36(26 ) & & & + & 0.934 & 3.54(29 ) & & & + & 2.639 & 10.05(30 ) & 10.60(27 ) & & + 2 & 0.342 & & & & + & 0.934 & 3.70(29 ) & & & + & 2.639 & 11.11(30 ) & 10.66(26 ) & &
we have performed a lattice simulation of a five - quark @xmath0@xmath1 like hadronic system on an anisotropic and asymmetric lattice .
the heavy quark anti - quark pair was treated as static .
thus it was possible to compute the total energy of the system as a function of the relative distance @xmath2 between the hadrons .
the maximum entropy method was employed toward this end .
an extrapolation into the physical pion mass region was also performed .
the objective was to extract an adiabatic potential @xmath3 for the relative motion .
the potential turns out to have two distinct attractive troughs , one at intermediate distances , @xmath183fm , and one at short range , @xmath149fm . to study the dynamics of the two - hadron system we have used @xmath3 in a schrdinger equation for three values of the physical reduced mass , corresponding to @xmath0@xmath1 , @xmath5@xmath6 , and @xmath7@xmath8 .
with increasing heavy - quark mass @xmath184 the number of s - wave bound states increases from one to four .
we have examined the corresponding wave functions to study the nature of the five - quark systems .
systematic errors predominantly originating from using the adiabatic approximation , extrapolation to zero pion mass , the non - relativistic framework , and possibly the quenched approximation , render the results of our study qualitative only .
nevertheless , we believe that the results for the @xmath7@xmath8 ground and excited state wave functions , respectively , bring to light a particularly interesting scenario with regard to five - quark system physics : the ground state is best described as a hadronic molecule , with a relative hadron - hadron distance matching a nuclear physics scale , whereas the excited state exhibits a wave function with support on very short distances of @xmath149fm , or so .
this leads us to interpret the excited state as a pentaquark .
the results for the @xmath11-quark system are less prone to relativistic corrections and therefore are our most reliable ones .
the light - quark @xmath0@xmath1 exhibits one ground state which clearly is a hadronic molecule . in the light of our results
it might be worthwhile to initiate future lattice qcd studies with this scenario in mind , but with a more realistic set of operators , lattices , actions , and lighter quark masses , etc .
all things considered , there is ground for the hypothesis that an all light - quark pentaquark may not exist , but that the existence of a genuine pentaquark in the heavy - quark sector can not be ruled out .
however , if present , it may well be an excited state . | we compute the ground - state energies of a heavy - light @xmath0@xmath1 like system as a function of the relative distance @xmath2 of the hadrons .
the heavy quarks , one in each hadron , are treated as static . then , the energies give rise to an adiabatic potential @xmath3 which we use to study the structure of the five - quark system . the simulation is based on an anisotropic and asymmetric lattice with wilson fermions .
energies are extracted from spectral density functions obtained with the maximum entropy method .
our results are meant to give qualitative insight : using the resulting adiabatic potential in a schrdinger equation produces bound state wave functions which indicate that the ground state of the five - quark system resembles a hadronic molecule , whereas the first excited state , having a very small rms radius , is probably better described as a five - quark cluster , or a pentaquark .
we hypothesize that an all light - quark pentaquark may not exist , but in the heavy - quark sector it might , albeit only as an excited state . |
we consider finite , undirected graphs @xmath7 with vertex set @xmath8 and edge set @xmath9 .
the graphs might have multiple edges but no loops .
let @xmath0 be a graph .
the length of the shortest ( odd ) cycle of the underlying simple graph of @xmath0 is called the _ ( odd ) girth _ of @xmath0 , i.e. the girth of @xmath0 is always at least three . for @xmath10
we denote by @xmath11 the set of edges with precisely one end in @xmath12 .
the minimum and maximum degree of @xmath0 is denoted by @xmath13 and @xmath14 , respectively .
a partial proper @xmath15-edge - coloring of @xmath0 is an assignment of colors @xmath16 to some edges of @xmath0 such that adjacent edges receive different colors .
let @xmath17 be a partial proper @xmath15-edge - coloring of @xmath0 .
the components of the subgraph which is induced by two colors @xmath18 and @xmath19 are called @xmath20-kempe - chains .
that is , a kempe - chain is either a path or an even cycle . in the case
that it is a path @xmath21 , we also say , that @xmath21 is an @xmath18-@xmath19-alternating path . a partial proper @xmath15-edge - coloring of @xmath0 is called a proper @xmath15-edge - coloring ( or just @xmath15-edge - coloring ) if all edges are assigned some color .
the smallest number @xmath22 for which @xmath0 has a @xmath22-edge - coloring is called the _ chromatic index _ of @xmath0 , and it is denoted by @xmath23 .
a graph @xmath0 is _ critical _ , if @xmath24 and @xmath25 , for every edge @xmath26 .
if @xmath0 is simple , we also say that @xmath0 is @xmath14_-critical_. clearly , @xmath27 , and by the classical theorems of shannon and vizing we have the following upper bounds for the chromatic index of a graph .
[ shannon](shannon ) let @xmath0 be a graph , then @xmath28 .
[ vizing](vizing ) let @xmath0 be a graph , then @xmath29 , where @xmath30 is the maximum multiplicity of an edge in @xmath0 .
a graph @xmath0 with @xmath31 is _ class i _ , otherwise it is _ class ii_. there are long standing open conjectures on class ii graphs , cf .
it is a notorious difficult open problem to characterize class ii graphs or even to obtain some insight into their structural properties .
this paper focuses on the @xmath3-edge - colorable part of graphs .
a subgraph @xmath2 of @xmath0 is called _ maximum _ @xmath3-edge - colorable , if it is @xmath3-edge - colorable and contains as many edges as possible .
the fraction @xmath32 is the subject of many papers , e.g. lower bounds are proved for cubic , subcubic or 4-regular graphs , @xcite .
one aim of this paper is to prove a general best possible lower bound for all graphs .
let @xmath2 be a maximum @xmath3-edge - colorable subgraph of @xmath0 , which is properly colored with colors @xmath33 .
usually , we will refer to edges of @xmath34 as uncolored edges . for
a vertex @xmath35 of @xmath0 let @xmath36 be the set of colors that appear at @xmath35 , and @xmath37 be the set of colors which are missing at @xmath35 .
let @xmath38 be an uncolored edge , @xmath39 , @xmath40 .
since @xmath2 is a maximum @xmath3-edge - colorable subgraph of @xmath0 , we have that @xmath41 and @xmath42 .
consider the @xmath18-@xmath19-alternating path @xmath21 starting from the vertex @xmath35 .
again , since @xmath2 is a maximum @xmath3-edge - colorable subgraph of @xmath0 , the path @xmath21 ends in @xmath43 .
thus @xmath21 is an even path , which together with the edge @xmath44 forms an odd cycle .
we will denote this cycle by @xmath45 .
if the subgraph @xmath2 is fixed , then we will shorten the notation to @xmath46 .
kempe chains forming an odd cycle together with an uncolored edge @xmath44 , @xmath47 , play a central role in the study of cubic graphs @xcite .
the second aim of this paper is to generalize some of these results to arbitrary graphs , and to investigate the maximum @xmath3-edge - colorable subgraphs .
we show that any set of vertex - disjoint cycles of a graph @xmath0 with @xmath48 can be extended to a maximum @xmath3-edge - colorable subgraph of @xmath0 .
in particular , any @xmath6-factor of a graph with maximum degree at least three can be extended to such a subgraph .
let @xmath49 be a @xmath23-edge - coloring of a graph @xmath0 with @xmath50 ( @xmath51 ) , and @xmath52 .
we define @xmath53 as the minimum size of the union of @xmath22 color - classes in a @xmath23-edge - coloring of @xmath0 .
let @xmath54 denote the minimum number of edges that should be removed from @xmath0 in order to obtain a graph @xmath2 with @xmath55 . clearly , @xmath56 , where @xmath2 is a maximum @xmath3-edge - colorable subgraph of @xmath0 .
in @xcite it is shown that the complement of any maximum @xmath57-edge - colorable subgraph of a cubic graph is a matching , and hence @xmath58 for cubic graphs .
this paper generalizes this result to simple graphs .
we further prove some bounds for the vertex degrees of a maximum @xmath14-edge - colorable subgraph @xmath2 .
the key property of cycles corresponding to uncolored edges in cubic graphs that is used in @xcite is their vertex - disjointness . as the example from figure [ fattriangleexample ] shows
, they can have even common edges in the general case . to see this ,
consider the graph @xmath0 and its maximum @xmath14-edge - colorable subgraph @xmath2 from figure [ fattriangleexample ] .
note that @xmath59 , @xmath60 , @xmath61 , and hence @xmath62 . + despite this , it turns out that , as theorem [ assignment ] demonstrates below , the edge - disjointness of the cycles can be preserved . for the proof
we need the following lemma , where we implicitly assume that the maximum @xmath14-edge - colorable subgraph @xmath2 of a graph @xmath0 is colored with @xmath63 colors .
before we proceed , let us note that if @xmath46 and @xmath64 are two cycles corresponding to two different uncolored edges @xmath44 and @xmath65 with respect to a maximum @xmath14-edge - colorable subgraph , then @xmath66 implies @xmath67 .
the lemma , that we are going to prove , describes the placement of the edges of @xmath44 and @xmath65 .
[ cyclesintersection ] let @xmath2 be any maximum @xmath14-edge - colorable subgraph of a graph @xmath0 , and @xmath44 , @xmath65 be two uncolored edges .
if @xmath68 , then there is a vertex @xmath35 such that @xmath44 and @xmath65 are incident to @xmath35 and @xmath69 .
let @xmath44 and @xmath65 be two uncolored edges with respect to @xmath2 , and @xmath70 . if @xmath71 , then @xmath44 and @xmath65 are parallel , therefore the statement of the lemma is trivial .
thus , we can assume that @xmath72 , and therefore every edge @xmath73 is colored with color @xmath18 .
we first show that @xmath44 and @xmath65 are adjacent .
assume to the contrary that this is not the case .
let @xmath74 be a subpath of @xmath75 that connects @xmath65 to @xmath76 , where @xmath77 is chosen so that @xmath78 .
note , that this is always possible .
suppose that @xmath74 connects @xmath65 to the vertex @xmath79 .
let @xmath21 be a subpath of @xmath46 that connects @xmath44 to the vertex @xmath35 , and does not pass through @xmath79 .
there are edges @xmath80 , @xmath81 which are adjacent to @xmath77 .
if @xmath82 , then @xmath83 is colored with color @xmath19 , and similarly , if @xmath84 , @xmath85 is colored with color @xmath86 . moreover , @xmath83 and @xmath85 do not share a vertex , @xmath83 is incident to @xmath35 , and @xmath85 to @xmath79 .
now recolor @xmath87 by leaving @xmath77 uncolored and color the remaining edges with colors @xmath18 , @xmath19 alternately , to obtain another maximum @xmath3-edge - colorable subgraph @xmath88of @xmath0 . by the choice of @xmath21 and @xmath74 , no edge of the subpath @xmath74 of @xmath75
is involved in the recoloring process .
thus @xmath89 can be colored with colors @xmath18 , @xmath86 and we obtain a @xmath3-edge - colorable subgraph @xmath90 with @xmath91 , contradicting our choice of @xmath2 . thus @xmath44 and @xmath65 are adjacent .
if @xmath44 and @xmath65 are parallel , then our statement follows easily . it remains to consider the case when @xmath44 and @xmath65 share precisely one vertex , say @xmath35 .
assume to the contrary that @xmath92 , i.e. @xmath41 .
then , interchanging colors @xmath18 and @xmath19 in @xmath46 allows us to color @xmath65 with color @xmath18 , contradicting the maximality of @xmath2 .
hence it follows that @xmath93 , and the statement is proved.@xmath94 [ assignment]let @xmath2 be any maximum @xmath14-edge - colorable subgraph of a graph @xmath0 , and let @xmath95 be the set of uncolored edges .
assume that @xmath2 is properly edge - colored with colors @xmath96 .
then there is an assignment of colors @xmath97 to the uncolored edges , such that @xmath98 , for all @xmath99 .
we prove the statement by induction on the number @xmath100 of uncolored edges . if @xmath101 , then the statement is trivial .
let @xmath102 , and @xmath0 be a graph with @xmath103 .
we need to consider two cases .
case 1 : there is a @xmath22 ( @xmath104 ) , such that @xmath105 .
then , @xmath2 is a maximum @xmath106-edge - colorable subgraph of a graph @xmath107 .
thus by the induction hypothesis , there are @xmath108 such that @xmath109 define @xmath110 , and @xmath111 . then @xmath112 , @xmath113 , and hence there are @xmath114 , such that @xmath115 and @xmath116 .
choose the colors @xmath117 to create @xmath118 , then lemma [ cyclesintersection ] implies that @xmath119 , for all @xmath99 .
case 2 : for each @xmath22 ( @xmath104 ) we have @xmath120 .
then , there is a vertex @xmath35 of maximum degree @xmath14 , such that @xmath35 is incident to all uncolored edges @xmath121 ( figure [ alluncoloreds ] ) .
thus , without loss of generality , we can assume that @xmath122 .
let @xmath123 ( @xmath124 ) be all the neighbors of @xmath35 such that @xmath35 and @xmath125 are connected by @xmath126 uncolored edges , @xmath127 , @xmath128 ( figure [ alluncoloreds ] )
. clearly , @xmath129 .
we may assume that the edges @xmath130 are incident to @xmath131 , the edges @xmath132 are incident to @xmath133 , ... , the edges @xmath134 are incident to @xmath135 .
moreover , let @xmath136 and @xmath137 .
.,title="fig : " ] + note that @xmath138 , for @xmath139 .
define @xmath140 now , it is not hard to see that the choice of @xmath141 and lemma [ cyclesintersection ] imply that @xmath142 , for all @xmath99 . @xmath94
the next theorem generalizes the result of @xcite that any @xmath6-factor of a cubic graph can be extended to a maximum @xmath57-edge - colorable subgraph to arbitrary graphs .
[ cyclesystemextension]let @xmath143 be any set of vertex - disjoint cycles of a graph @xmath0 with @xmath144 . then there is a maximum @xmath3-edge - colorable subgraph @xmath2 of @xmath0 , such that @xmath145 .
let @xmath144 . for @xmath143 consider a maximum @xmath3-edge - colorable subgraph @xmath2 of @xmath0 with @xmath146 . we will show that @xmath145 .
assume to the contrary that there is an edge @xmath147 , such that @xmath44 does not belong to @xmath148 .
let us assume that @xmath2 is properly colored with colors @xmath149 .
case 1 : there are @xmath150 and @xmath151 , such that @xmath46 contains an edge @xmath83 that does not belong to a cycle of @xmath143 . consider a proper partial @xmath3-edge - coloring of @xmath0 obtained from the coloring of @xmath2 by shifting the colors on the cycle @xmath46 and leaving the edge @xmath83 uncolored .
the new partial @xmath3-edge - coloring corresponds to a maximum @xmath3-edge - colorable subgraph @xmath88 of @xmath0 with @xmath152 , contradicting the choice of @xmath2 .
case 2 : for all @xmath150 and @xmath151 , the edges of @xmath46 lie on a cycle of @xmath143 .
since the cycles of @xmath143 are vertex - disjoint , it follows that @xmath153 and there is only one cycle @xmath154 , which , in its turn , is a cycle of @xmath143 .
this implies that @xmath155 and @xmath156 .
let us assume that @xmath157 is colored by the colors @xmath158 and @xmath3 , alternately .
for an edge @xmath83 of @xmath159 , define @xmath160 obtained from @xmath2 as follows , 1 .
leave the edge @xmath83 uncolored , 2 .
color the edges of the even path @xmath161 with colors @xmath158 and @xmath3 , alternately , 3 .
leave the colors of the rest of edges unchanged .
note that the subgraphs @xmath162 and @xmath2 are the same , though they may have received different colorings , and for each edge @xmath83 of @xmath159 , @xmath160 is a maximum @xmath3-edge - colorable subgraph of @xmath0 with @xmath163 .
similar to the consideration of the case 1 with respect to @xmath160 , it can be shown that we can assume that if @xmath164 is an edge of @xmath159 , then @xmath165 and @xmath166 .
this implies that the vertices of the cycle @xmath159 are of degree @xmath3 . since @xmath5 ,
each vertex of @xmath159 is incident to @xmath167 edges that do not lie on @xmath168 , and which are colored by the colors @xmath169 in every @xmath160 .
let @xmath17 be a proper partial @xmath3-edge - coloring of @xmath0 which is obtained from the coloring of @xmath2 by deleting the colors of all edges of @xmath168 .
since @xmath168 is an odd cycle , it follows that there is a @xmath170-@xmath3-alternating path @xmath171 ( with respect to the edge - coloring @xmath17 ) that starts from a vertex @xmath172 and does not end at a vertex of @xmath168 .
choose an edge @xmath173 , and let @xmath174 be the other edge of @xmath168 that is incident to @xmath79 .
consider a proper partial @xmath3-edge - coloring of @xmath0 obtained from @xmath17 as follows , 1 .
shift the colors on the path @xmath171 , and clear the color of the edge that is incident to @xmath175 and has a color @xmath170 in @xmath17 , 2 .
color @xmath176 with color @xmath170 , and color the edges of the even path @xmath177 with colors @xmath158 and @xmath3 alternately , starting from the edge @xmath174 .
this new partial @xmath3-edge - coloring of @xmath0 corresponds to a maximum @xmath3-edge - colorable subgraph @xmath88 of @xmath0 , which satisfies latexmath:[$ contradicting the choice of @xmath2 .
@xmath94 [ 2factorextension]let @xmath143 be any 2-factor of a graph @xmath0 with @xmath48 .
then , there is a maximum @xmath14-edge - colorable subgraph @xmath2 of @xmath0 , such that @xmath145 .
[ cutcondition ] let @xmath2 be any maximum @xmath14-edge - colorable subgraph of a graph @xmath0
. then 1 .
@xmath179 for each @xmath10 , 2 .
@xmath180 for each vertex @xmath181 of @xmath0 , and 3 .
. furthermore , the bounds are best possible .
let @xmath10 , and assume that @xmath11 contains @xmath22 uncolored edges . by theorem [ assignment ] , there is an assignment of colors to uncolored edges with respect to @xmath2 such that the corresponding cycles do not intersect .
since every cycle @xmath46 of an uncolored edge @xmath44 of @xmath11 , intersects @xmath11 at least twice , there are at least @xmath22 pairwise different edges of @xmath2 that belong to @xmath11 . statements 2 . and 3 .
follow directly from the first .
it remains to show that the bounds are best possible .
let @xmath183 be the complete bipartite graph @xmath184 with a subdivided edge .
it is well known , that @xmath183 is @xmath185-critical .
therefore it has a @xmath185-edge - coloring , that leaves precisely one edge , which is incident to a vertex of degree 2 , uncolored .
take @xmath186 copies of @xmath183 and identify the vertices of degree two to obtain the graph @xmath2 .
now take two copies of @xmath2 and connect the vertices of degree @xmath187 with an edge to obtain the graph @xmath0 .
it is not hard to observe that @xmath0 has a maximum @xmath185-edge - colorable subgraph of minimum degree @xmath188 .
this implies that the bounds are best possible ( for statement 3 as well , since @xmath0 is regular ) .
note , that the graphs are simple .
for multi - graphs , take @xmath189 copies of the graph with three vertices , one vertex of degree 2 and the other two of degree @xmath190 .
these graphs have a @xmath190-edge - colorable subgraph that leaves precisely one edge uncolored , which is incident to a bivalent vertex .
identify the bivalent vertices to obtain a @xmath190-regular graph with the desired properties .
next we will construct some graphs to which we will refer in the following to show that the bounds of theorem [ multigraphgirth ] are best possible .
recall that , by definition , the girth of a graph is always at least three .
[ example ] for each @xmath191 , there is a graph @xmath192 with girth @xmath193 , such that @xmath194 , for each maximum @xmath195-edge - colorable subgraph @xmath196 .
furthermore , there is a maximum @xmath195-edge - colorable subgraph @xmath197 with @xmath198 . for @xmath191 ,
let @xmath199 be a cycle of length @xmath200 , where each edge has multiplicity @xmath190 .
then @xmath201 , and @xmath202 .
every color class contains at most @xmath22 edges , hence @xmath203 , for every maximum @xmath3-edge - colorable subgraph .
let @xmath197 be the subgraph of @xmath192 with @xmath204 edges , one edge of multiplicity @xmath190 and the remaining @xmath190 edges of multiplicity @xmath205 .
it is easy to see that @xmath197 is @xmath206-edge - colorable , thus @xmath207 for all maximum @xmath3-edge - colorable subgraphs of @xmath192 .
. @xmath94 [ multigraphgirth ] if @xmath0 is a graph with girth @xmath209 ( @xmath210 ) , and @xmath2 a maximum @xmath14-edge - colorable subgraph of @xmath0 , then @xmath211 , and the bound is best possible .
let @xmath2 be a maximum @xmath14-edge - colorable subgraph of @xmath0 , and @xmath212 be the set of uncolored edges .
let @xmath213 be the pairwise edge - disjoint cycles of theorem [ assignment ] , @xmath214 , and @xmath88 the colored subgraph of @xmath215 .
then it follows by theorem [ assignment ] , that @xmath216 .
furthermore , @xmath217 , and hence @xmath218 .
the graphs @xmath192 of lemma [ example ] show that the bound is best possible .
@xmath94 theorem [ cutcondition ] implies that @xmath219 .
however for the maximum vertex degree of a maximum @xmath14-edge - colorable subgraph @xmath2 of @xmath0 much better bounds can be proved . for this
we need the well known fact , that ( * ) @xmath220 for graphs @xmath0 with @xmath221 and odd girth @xmath176 .
this result is easily deducible from the results of kierstead @xcite ( about acceptable paths ) , and it implies that @xmath222 . in @xcite it is proved that @xmath223 .
we summarize the consequences of these two results in the following corollary . if @xmath0 is a graph with girth @xmath209 @xmath224 , and @xmath2 a maximum @xmath14-edge - colorable subgraph of @xmath0 , then 1 .
@xmath225 , 2 .
@xmath226 .
[ matchingcomplement ] every simple graph @xmath0 contains a maximum @xmath3-edge - colorable subgraph , such that the uncolored edges form a matching .
take a maximum @xmath3-edge - colorable subgraph @xmath2 of @xmath0 which minimizes the number of pairs of adjacent uncolored edges .
the proof will be completed if we show that this number is zero .
assume , on the contrary , that there is a vertex @xmath35 which is incident to two uncolored edges , one of which is @xmath227 .
let @xmath2 be @xmath3-edge - colored and assume that @xmath228 , @xmath229 .
since @xmath2 is a maximum @xmath3-edge - colorable subgraph of @xmath0 , there is an edge @xmath230 that is colored with color @xmath231 .
consider a maximal fan beginning from the vertex @xmath232 , that is a maximal sequence @xmath233 such that 1 .
@xmath234 , and for @xmath235 , the edge @xmath236 is colored with color @xmath237 , 2 .
@xmath238 , for @xmath239 , and 3 .
@xmath240 are distinct colors .
we show that @xmath241 .
suppose that @xmath242 .
this means that @xmath243 is a vertex of maximum degree @xmath14 and all edges incident to it are from @xmath2 .
consider a maximum @xmath3-edge - colorable subgraph @xmath88 of @xmath0 obtained from @xmath2 as follows : color @xmath227 by @xmath231 , for @xmath244 color @xmath236 with color @xmath245 , and leave the edge @xmath246 uncolored .
then there are less pairs of adjacent uncolored edges in @xmath88 than in @xmath2 , which contradicts the choice of @xmath2 .
thus , there exist @xmath247 .
the maximality of the sequence @xmath248 implies that @xmath249 , say @xmath250 , for an @xmath251 .
recall that @xmath252 .
the maximality of @xmath2 implies that @xmath253 .
consider the maximal @xmath19-@xmath254-@xmath237 alternating path @xmath21 beginning from the vertex @xmath243 .
we will show that our assumptions imply that @xmath2 is not a maximum @xmath3-edge - colorable subgraph of @xmath0 , contradicting the choice of @xmath2 .
case 1 : @xmath21 does not reach @xmath255 .
the following recoloring yields the desired contradiction .
color the edge @xmath227 by @xmath231 , @xmath256 by @xmath257 , @xmath258 , exchange the colors on the path @xmath21 and color @xmath246 by @xmath19 .
case 2 : @xmath21 reaches @xmath259 . since the edge @xmath236 is colored by @xmath237 it follows that the path @xmath21 enters @xmath259 by an edge colored @xmath19 . to get the contradiction , color the edge @xmath227 by @xmath231 , @xmath256 by @xmath257 , @xmath260 , exchange the colors on the path @xmath21 and color @xmath236 by @xmath19 .
case 3 : @xmath21 reaches @xmath261 . since @xmath238 , it follows that the path @xmath21 reaches @xmath261 by an edge colored @xmath19 .
color the edge @xmath227 by @xmath231 , @xmath256 by @xmath257 , @xmath262 , exchange the colors on the path @xmath21 and color @xmath263 by @xmath19 to get the desired contradiction .
@xmath94 theorem [ matchingcomplement ] is equivalent to if @xmath0 is a simple graph , then @xmath58 .
lemma [ example ] shows that a maximum @xmath3-edge - colorable subgraph of a multigraph can be class ii as well .
this can not be the case for simple graphs as the following theorem shows .
[ equality ] if @xmath2 is a maximum @xmath14-edge - colorable subgraph of a simple graph @xmath0 , then @xmath264 , i.e. @xmath2 is class i. let @xmath265 be an uncolored edge
. then @xmath266 , and hence , since @xmath267 is simple , @xmath268 . the graph @xmath267 contains a @xmath14-critical subgraph @xmath88 , which clearly contains the edge @xmath44 . by vizing s adjacency lemma @xcite
, every vertex of @xmath88 is adjacent to at least two vertices of maximum degree @xmath14 .
thus there is a vertex @xmath269 of @xmath88 with @xmath270 , and hence @xmath271 .
theorem [ matchingcomplement ] says , that every simple class ii graph @xmath0 has a maximum @xmath3-edge - colorable subgraph @xmath2 , such that @xmath272 .
we believe that this can be generalized to multigraphs .
we would like to thank our reviewers for their useful comments that helped us to improve the presentation of the paper .
our special thanks to the second reviewer , who has pointed out a mistake in the earlier version of the proof of lemma [ cyclesintersection ] .
d. aslanyan , v. v. mkrtchyan , s. s. petrosyan , g. n. vardanyan , on disjoint matchings in cubic graphs : maximum @xmath6- and @xmath57-edge - colorable subgraphs , under review , ( available at : http://arxiv.org/abs/0909.2767 ) | * abstract : * a graph @xmath0 is class ii , if its chromatic index is at least @xmath1 . let @xmath2 be a maximum @xmath3-edge - colorable subgraph of @xmath0 .
the paper proves best possible lower bounds for @xmath4 , and structural properties of maximum @xmath3-edge - colorable subgraphs .
it is shown that every set of vertex - disjoint cycles of a class ii graph with @xmath5 can be extended to a maximum @xmath3-edge - colorable subgraph .
simple graphs have a maximum @xmath3-edge - colorable subgraph such that the complement is a matching .
furthermore , a maximum @xmath3-edge - colorable subgraph of a simple graph is always class i. + * keywords : * maximum @xmath3-edge - colorable subgraph , matching , @xmath6-factor , edge - chromatic number , chromatic index , class ii graph |
in the great number of topological insulators ( ti ) and topological superconductors ( tsc ) discovered or proposed@xcite , the transition from topologically trivial to nontrivial phases can in general be triggered by varying all kinds of energy parameters , e.g. , chemical potential@xcite , hopping@xcite , interface coupling@xcite , etc .
the canonical way of identifying the topological phase transition is to first consult the dimension and symmetry class@xcite of the system to see if topologically nontrivial phases are possible , and if so then seek for the value of the driving energy parameter at which the spectral gap closes at certain momenta , since topological phase transition necessarily involves band inversion .
ideally , the topological invariant jumps discretely when the system undergoes a topological phase transition , therefore it seems rather ambiguous to identify any asymptotic critical behavior from the topological invariant itself .
drawing analogy from the landau order parameter paradigm , a profound question is whether there exists a universal scaling scheme that can judge the topological phase transitions driven by any energy parameter in any dimension and symmetry class , and if the scaling procedure is by any means different from kadanoff s scaling theory@xcite since the systems under question possess no landau order parameter .
moreover , it would be of tremendous usage if the scaling scheme also renormalizes the driving energy parameter . the proposal related to entanglement
entropy first shed a light on this issue@xcite .
on the other hand , since a great number of tis and tscs are characterized by the topological invariants that are often calculated from the momentum space integration of a certain integrand function that represents the curvature of the many - body state , it is natural to suspect that a scaling scheme that renormalizes the integrand function can be used to judge topology . depending on the system , the integrand function may be berry curvature@xcite , berry connection@xcite , or any other quantity whose integration gives the topological invariant .
for the sake of a general discussion , hereafter the integrand function is referred to as the curvature function , and the topological invariant calculated from the integration of the curvature function is referred to as the winding number ( to draw analogy to the number of knots in a closed string , and to distinguish it from other means of calculating topological invariants ) . if a scaling scheme for the curvature function exists and gives a rg flow of the driving energy parameter that judges the topological phase transition , the following criteria are expected : ( 1 ) the fixed point in the parameter space of the driving energy parameter has a particular configuration of curvature function that is invariant under the scaling procedure .
( 2 ) any point in the parameter space flows into some fixed point while preserving the winding number during the flow .
( 3 ) different fixed points correspond to different winding numbers . in this article
, we propose a scaling procedure that satisfies the aforementioned criteria for inversion - symmetric systems .
the motivation comes from a simple observation : to know the number of knots that a messy string contains , one can either integrate the curvature along the string ( integrating the curvature function to get winding number ) , or simply stretch the string until the knots become obvious ( the proposed scaling procedure ) .
a single equation is proposed to implement this knot - tying principle in ti and tsc : let @xmath0 be the curvature function at momentum @xmath1 from which the winding number in @xmath2-dimension @xmath3 is calculated , with @xmath4 the driving energy parameter that controls the topology . given a @xmath4 in the parameter space , we seek for a new @xmath5 that satisfies @xmath6 where @xmath7 is a high symmetry point , and @xmath8 is a small deviation away from it satisfying @xmath9 . as we shall see in the examples below , the operation of eq .
( [ scaling_scheme_general ] ) in inversion - symmetric ti or tsc correctly captures the topological phase transition driven by any energy parameter .
the fixed point is reached by iteratively solving @xmath4 , and by considering a small deviation @xmath8 , one obtains the rg equation for @xmath4 in the form of differential equation .
the curvature function gradually evolves into a fixed point configuration that is invariant under this procedure , analogous to a string with its knots tight and can not be stretched anymore . moreover , for a great variety of models , an asymptotic universal critical behavior of the curvature function near the gap - closing momenta is revealed despite the system in general has no landau order parameter or any short range correlation .
the critical behavior may be varified by ultracold atoms in optical lattices in the case that the curvature function is the berry curvature .
to explain the mechanism behind eq .
( [ scaling_scheme_general ] ) and demonstrate its analogy to knot - tying , consider a inversion - symmetric ti or tsc defined on a @xmath2-dimensional cubic lattice . if the system at @xmath4 and at the fixed point @xmath10 have the same topology , then the curvature function at @xmath4 can be expanded by @xmath11 where @xmath12 is the fixed point curvature function that satisfies @xmath13 , i.e. , it is invariant under the operation of eq .
( [ scaling_scheme_general ] ) , and @xmath14 is the deviation away from this fixed point configuration at @xmath4 , which has fourier component @xmath15 .
the expansion of @xmath14 is sound because it must not contribute to the integration of winding number @xmath16 such that the system at @xmath4 and at @xmath10 have the same topology .
suppose we choose a particular high symmetry point such as @xmath17 , and a small displacement away from it along @xmath18-th coordinate @xmath19 .
now we apply the operation of eq .
( [ scaling_scheme_general ] ) by using eq .
( [ berry_curvature_d_dimension ] ) .
expanding around @xmath7 gives @xmath20 where @xmath21 is the sign of @xmath22 at the chosen @xmath7 and at the fourier component @xmath23 .
note that @xmath24 is the deviation at @xmath7 .
writting @xmath25 and introducing a pseudo scaling parameter @xmath26 , one obtains @xmath27 this means the deviation at @xmath7 changes under this operation according to the slope of the deviation at @xmath28 .
if the initial value @xmath4 gives @xmath29 , i.e. , a positive ( negative ) deviation at @xmath7 , then @xmath14 must curve down ( curve up ) as moving from @xmath7 to @xmath28 in order to conserve the winding number , so @xmath30 and consequently @xmath31 is reduced under this scaling procedure .
thus continuously applying eq .
( [ scaling_scheme_general ] ) makes @xmath14 approaching zero and @xmath0 approaching @xmath12 . in other words , the principle behind eq .
( [ scaling_scheme_general ] ) is that the deviation of the curvature function from its fixed point configuration is gradually reduced under this scaling procedure .
as we shall see below , in @xmath32 systems , this is synonymous to stretching a messy string to reduce its messiness until the knots are tight .
in general , there could be a situation that @xmath33 and @xmath34 at a given @xmath4 so @xmath31 increases at the beginning of the scaling process , but @xmath31 must eventually be reduced because the scaling process still leads to @xmath35 in order to conserve winding number .
expanding eq .
( [ scaling_scheme_general ] ) in both @xmath36 and @xmath37 yields the rg equation . in the leading order ,
@xmath38 this explains that the proper scaling parameter should be defined as quadratic in the displacement @xmath37 , whereas the @xmath39 in eq .
( [ rg_equation_d_dimension ] ) is merely introduced to demonstrate the deviation - reduction mechanism .
since the rg equation involves only one variable @xmath4 , it is always possible to map the rg flow into the motion of an overdamped particle in a conservative potential@xcite .
equation ( [ berry_curvature_d_dimension ] ) , together with the fact that at critical point @xmath40 the curvature function diverges at the gap - closing momentum which is at a high symmetry point @xmath7 if the system is inversion - symmetric@xcite , indicate that the curvature function near the gap - closing point @xmath7 can be expanded by @xmath41 as shown below , @xmath42 explicitly characterizes the scale invariance at the fixed point or critical point .
note that although eq .
( [ correlation_length_definition ] ) is similar to the ornstein - zernike form of correlation function@xcite , @xmath42 does not represent the correlation length of short range fluctuations which is obviously absent in ti and tsc .
instead , it is a length scale defined from the curvature function near the gap - closing point that signals the scale invariance under the proposed scaling scheme .
alternatively , one may use its inverse @xmath43 as a momentum scale to characterize the scale invariance , which works equally well .
two remarks are made before we move to concrete examples .
firstly , as in any rg procedure , this scaling scheme itself does not give a meaning to the fixed point , i.e. , it does not tell us whether the fixed point is topologically trivial or nontrivial , which may nevertheless be clarified by direct calculation of eq . ( [ winding_number_integration ] ) at @xmath10 or at any @xmath4 that flows to the @xmath10 .
secondly , @xmath4 may increase or decrease along the rg flow , but the notion of relevant , irrelevant , or marginal coupling constants@xcite does not seem to apply to @xmath4 , since the topological phase transitions are judged by the sign change of the rg equation in eq .
( [ generic_rg_equation ] ) which is not directly related to how @xmath4 changes under the scaling procedure . ) , applied to the topologically nontrivial phase of the ssh model with the choice @xmath44 . given an initial @xmath45 and the corresponding berry connection @xmath46 ( red line ) , we find a new @xmath47 by demanding @xmath46 at @xmath48 to be equal to that of @xmath49 ( orange line ) at @xmath50 , as indicated by the dash line .
this procedure reduces the deviation away from and leads to the fixed point configuration @xmath51 ( blue line ) .
the @xmath52 expressed as a vector field in the complex space for each configuration is indicated by colored arrows .
( b ) the length scale @xmath42 defined in eq .
( [ correlation_length_definition ] ) as a function of @xmath45 .
( c ) the rg flow of @xmath45 .
when joining the @xmath52 arrows in ( a ) at a particular @xmath45 head to tail in sequence to form a string , this scaling procedure resembles stretching the string to reveal whether it has a knot , as indicated by color arrows .
, title="fig : " ] ) , applied to the topologically nontrivial phase of the ssh model with the choice @xmath44 .
given an initial @xmath45 and the corresponding berry connection @xmath46 ( red line ) , we find a new @xmath47 by demanding @xmath46 at @xmath48 to be equal to that of @xmath49 ( orange line ) at @xmath50 , as indicated by the dash line .
this procedure reduces the deviation away from and leads to the fixed point configuration @xmath51 ( blue line ) .
the @xmath52 expressed as a vector field in the complex space for each configuration is indicated by colored arrows .
( b ) the length scale @xmath42 defined in eq .
( [ correlation_length_definition ] ) as a function of @xmath45 .
( c ) the rg flow of @xmath45 .
when joining the @xmath52 arrows in ( a ) at a particular @xmath45 head to tail in sequence to form a string , this scaling procedure resembles stretching the string to reveal whether it has a knot , as indicated by color arrows .
, title="fig : " ]
to demonstrate the scaling scheme in @xmath32 , we start from the spinless ssh model@xcite with periodic boundary condition ( pbc ) , described by the hamiltonian @xmath53 the topology of this model is judged by the winding number @xmath54 of the operator @xmath55 in the complex space when one goes through the entire brillouin zone ( bz ) . defining @xmath56 ,
the winding number is calculated from the curvature function which is the berry connection@xcite @xmath46 in this case , @xmath57 we now search for a new @xmath58 by applying eq .
( [ scaling_scheme_general ] ) @xmath59 writing @xmath60 , @xmath61 , and using @xmath62 one obtains the rg equation at the leading order @xmath63 both reproduce the correct critical points at @xmath64 and fixed points at @xmath65 , as indicated by the rg flow in fig .
[ fig : ssh_model_scaling ] . through directly calculating the winding number ,
the @xmath66 fixed point is topologically trivial and the @xmath67 fixed point is nontrivial .
the example shown in fig .
[ fig : ssh_model_scaling ] ( a ) ( red line ) clearly demonstrates the deviation - reduction mechanism of eq .
( [ rg_equation_d_dimension ] ) , where the initial value @xmath68 gives @xmath69 and @xmath70 such that @xmath71 is gradually reduced to zero under the operation of eq .
( [ scaling_scheme_general ] ) .
the choice of different @xmath50 means different @xmath71 to start with , hence the speed of converging to the fixed point configuration is also different , as reflected in the two equations in eq .
( [ rg_eq_ssh ] ) .
note that the phase transition occurs when the gap at @xmath44 closes , but the scaling scheme works in this model even if one chooses @xmath72 that is not the gap - closing point .
alternatively , one can introduce the ratio @xmath73 and discuss the topological phase transition upon tuning @xmath74 . using eq .
( [ scaling_scheme_general ] ) with @xmath72 leads to @xmath75 which well reproduces the two critical points at @xmath76 .
thus the proposed scaling scheme is valid whether the system is parametrized by @xmath45 or @xmath74 . from eq .
( [ correlation_length_definition ] ) we obtain the @xmath42 deduced from the gap - closing point @xmath44 @xmath77 as shown in fig . [ fig : ssh_model_scaling ] , @xmath78 at the critical point @xmath79 , and @xmath80 at the topologically nontrivial fixed point @xmath81 , both signature the scale invariance@xcite .
we emphasize that the scaling scheme here is a procedure akin to knot - tying , and scale - invariance means that the curvature function converges to a configuration analogous to a string with all its knots tight ( curvature function at gap - closing point stops changing ) , as indicated in fig .
[ fig : ssh_model_scaling ] ( c ) , both have different meaning than those in kadanoff s scaling theory .
the topologically trivial fixed point @xmath82 has @xmath83 , meaning that while the amplitude of @xmath46 is approaching zero everywhere , its functional form is approaching the first harmonic @xmath84 .
chern insulators in the continuum , which has the form of eq .
( [ chern_berry_generic_form ] ) , at evenly spaced values of @xmath85 .
all the lines evolve to the fixed point configuration labeled by the blue line under the proposed scaling procedure .
( b ) the length scale @xmath42 defined in eq .
( [ correlation_length_definition ] ) diverges at @xmath86 and vanishes at @xmath87 .
we set @xmath88 in these plots .
( c ) the rg flow of @xmath89 . ,
title="fig : " ] chern insulators in the continuum , which has the form of eq .
( [ chern_berry_generic_form ] ) , at evenly spaced values of @xmath85 .
all the lines evolve to the fixed point configuration labeled by the blue line under the proposed scaling procedure .
( b ) the length scale @xmath42 defined in eq .
( [ correlation_length_definition ] ) diverges at @xmath86 and vanishes at @xmath87 .
we set @xmath88 in these plots .
( c ) the rg flow of @xmath89 .
, title="fig : " ] to demonstrate the feasibility of eq .
( [ scaling_scheme_general ] ) in higher dimensions , we consider spinless chern insulators in the @xmath2-dimensional continuum@xcite , the low energy sector of which has the generic dirac form @xmath90 where @xmath91 satisfy the clifford algebra .
the curvature function from which the winding number is calculated , take @xmath92 and @xmath93 as examples , is @xmath94 which has the generic form @xmath95^{\frac{d+1}{2 } } } \label{chern_berry_generic_form}\end{aligned}\ ] ] where the prefactor @xmath96 depends on symmetry and dimension but is unimportant for our argument .
( [ scaling_scheme_general ] ) with the only high symmetry point @xmath97 and @xmath98 , writing the energy parameter to be renormalized as @xmath99 while keeping @xmath100 constant , and defining @xmath101 , the leading order rg equation is @xmath102 which has generically a critical point at @xmath86 , and the two fixed points @xmath87 and @xmath103 assuming @xmath104 .
the length scale calculated from eq .
( [ correlation_length_definition ] ) @xmath105 vanishes at the fixed points and diverges at the critical point .
the results for @xmath92 spinless chern insulators , which have @xmath106 , @xmath107 , and @xmath108 are shown in fig .
[ fig : chern_continuum_scalin ] , where the curvature function is the berry curvature@xcite .
in contrast to the knot - tying picture in @xmath32 , the scaling procedure in @xmath92 is a process to stretch the skyrmion texture of @xmath109 in eq .
( [ dirac_hamiltonian ] ) without changing the skyrmion number@xcite until the curvature function flattens to second order at the gap - closing point ( blue line in fig .
[ fig : chern_continuum_scalin](a ) ) .
two systems of similar kind are haldane s @xmath92 graphene model@xcite and kane - mele model@xcite . consider the spinless haldane model whose expansion around @xmath110 and @xmath111 points of the reciprocal space of the hexagonal lattice
is described by the hamiltonian@xcite @xmath112 where upper sign is for @xmath113 and the lower sign @xmath114 . applying eqs .
( [ chern_berry_general_generic_form ] ) and ( [ scaling_scheme_general ] ) yields @xmath115 correctly reproducing the critical points at @xmath116 . at high symmetry points for the four phases of @xmath92 chern insulators .
( b ) the rg flow of @xmath89 by choosing three different high symmetry points @xmath7 with the same @xmath117 .
one sees that choosing a particular @xmath7 captures the transition caused by gap - closing at this point , in accordance with the sign change of @xmath118 in ( a ) .
, title="fig : " ] at high symmetry points for the four phases of @xmath92 chern insulators .
( b ) the rg flow of @xmath89 by choosing three different high symmetry points @xmath7 with the same @xmath117 .
one sees that choosing a particular @xmath7 captures the transition caused by gap - closing at this point , in accordance with the sign change of @xmath118 in ( a ) .
, title="fig : " ] to discuss the chern insulators on a @xmath2-dimensional cubic lattice , we consider eq .
( [ dirac_hamiltonian ] ) in @xmath92 and @xmath93 with components@xcite @xmath119 the @xmath120 high symmetry points in the first quartet @xmath121 consist of each @xmath122 being either @xmath123 or @xmath124 .
the curvature function expanded around each @xmath7 takes the form @xmath125}{\left\{|\delta{\bf
k}|^{2}+\left[\left(m-4bn_{\pi}\right)-\xi(\delta{\bf k})\right]^{2}\right\}^{\frac{d+1}{2}}}\ ; , \nonumber \\
\xi(\delta{\bf k})&=&b\left(\sum_{i\in 0}\delta k_{i}^{2}-\sum_{i\in \pi}\delta k_{i}^{2}\right)\end{aligned}\ ] ] where @xmath126 ( @xmath127 ) denotes summation over @xmath128 at which @xmath129 ( @xmath130 ) , and @xmath131 is the number of @xmath124 s in @xmath132 .
applying eq .
( [ scaling_scheme_general ] ) with @xmath133 along one particular coordinate @xmath134 yields @xmath135 and hence the generic fixed points @xmath136 and @xmath137 when choosing a particular @xmath7 .
the length scale @xmath42 defined from the gap - closing point satisfies @xmath138 the top sign in each @xmath139 or @xmath140 corresponds to the case when the component of @xmath7 in the scaling direction @xmath133 is @xmath141 , and the bottom sign is when @xmath142 . because @xmath131 takes any integer value from @xmath123 to @xmath2 , generically there are @xmath143 critical points located at @xmath144 .
the phase transition at a particular @xmath145 takes place when the gap closes@xcite at one set of @xmath7 s that have the same @xmath131 , in accordance with applying eq .
( [ scaling_scheme_general ] ) at these @xmath7 s to capture the change of curvature function near them , as shown for @xmath92 in fig .
[ fig:2d_chern_d3 ] .
the chiral @xmath146-wave superconductors in the continuum and in the lattice@xcite have the same generic form of curvature function as the chern insulators@xcite ( with the replacement @xmath147 chemical potential and @xmath148effective mass ) , and hence practically the same critical points and critical behavior .
we remark that the ssh model@xcite and haldane s graphene model@xcite have been realized by ultracold atoms in optical lattices , where the systems can be driven close to the critical point .
the predicted critical behavior of @xmath42 can be verified in the systems where the curvature function is the berry curvature , which may be measurable by detecting the anomalous velocity associated with the berry curvature@xcite or momentum space interferometry techniques@xcite .
in summary , we present a scaling procedure to judge topological phase transitions driven by any energy parameter . the procedure is valid for inversion - symmetric models in any dimension and symmetry class provided the topological invariant is calculated from the integration of a certain curvature function .
our formalism reveals that the concept of scaling in topologically ordered systems falls into a completely different realm than kadanoff s scaling theory in the landau order parameter paradigm . based on a simple knot - tying picture , the scaling procedure renormalizes the curvature function while keeping the winding number intact , through which the rg flow of the driving energy parameter @xmath4 is obtained , as well as the rg equation , eq .
( [ generic_rg_equation ] ) , under an infinitesimal operation .
in essence , the scaling procedure uses the divergence of second derivative of the curvature function at the gap - closing momentum to find the critical point , and uses the flattening of the second derivative to find the fixed point , both achieved under a single operation of eq .
( [ scaling_scheme_general ] ) .
a length scale defined from the berry curvature near the gap - closing momentum shows an asymptotic universal critical behavior that diverges at the critical point in first power and vanishes at the nontrivial fixed point in square root if it is finite @xmath149 for a variety of inversion - symmetric systems examined .
applications to a broader class of models , such as those driven by interactions or inversion - asymmetric models , will be subject to future investigations . | topologically ordered systems are characterized by topological invariants that are often calculated from the momentum space integration of a certain function that represents the curvature of the many - body state .
the curvature function may be berry curvature , berry connection , or other quantities depending on the system .
akin to stretching a messy string to reveal the number of knots it contains , a scaling procedure is proposed for the curvature function in inversion symmetric systems , from which the topological phase transition can be identified from the flow of the driving energy parameters that control the topology ( hopping , chemical potential , etc . ) under scaling . at an infinitesimal operation , one obtains the renormalization group ( rg ) equations for the driving energy parameters .
a length scale defined from the curvature function near the gap - closing momentum is suggested to characterize the scale invariance at critical points and fixed points , and displays a universal critical behavior in a variety of systems examined . |
this appendix presents additional numerical evidence which demonstrates that the results of exact diagonalization ( ed ) presented in fig .
5 for the two models considered in the letter are robust . in each case
the spectra only weakly depend on the model parameters in the regime of the zigzag magnetic phase . for convenience , we repeat first the definitions of the models considered in the ed calculations .
we consider the following @xmath0-@xmath1-like _ model i _
, @xmath131 on the honeycomb lattice .
the operators @xmath132 in the hopping term @xmath28 , with @xmath26 , stand for creation of composite fermions with pseudospin flavor @xmath29 in the restricted space without double occupancies ; the constraint follows from large on - site coulomb repulsion @xmath32 .
the exchange term @xmath33 describes in this case the heisenberg - kitaev model , with ferromagnetic ( fm ) heisenberg ( @xmath34 ) and antiferromagnetic ( af ) kitaev ( @xmath35 ) exchange , given by the interaction @xmath1 and @xmath37 $ ] , see eqs .
( [ jhk ] ) .
the parameter @xmath40 interpolates between the heisenberg ( @xmath133 ) and kitaev ( @xmath134 ) exchange couplings acting between pairs of nearest neighbor ( nn ) spins @xmath6 ; zigzag af order is found in a broad range of @xmath40 @xcite .
the second model investigated in the letter , called _ model ii _ , has inverted signs of the nn spin couplings in the @xmath41 term , and contains in addition exchange interactions between next nn ( nnn ) and third nn ( 3nn ) spins , @xmath42 with @xmath135 @xcite : @xmath136 further neighbor heisenberg interactions @xmath137 are necessary to stabilize zigzag af order which is unstable in case of only nn af heisenberg and fm kitaev interactions ( at @xmath138 ) @xcite . as obtained by ed for a periodic cluster of @xmath9 sites at selected high - symmetry points @xmath18 , @xmath25 , @xmath24 , and @xmath104 in the bz at @xmath108 and @xmath69 for model i eq .
( 1 ) , with : ( a ) @xmath139 , and ( b ) @xmath67 . ,
width=302 ] the ed calculations were performed on a periodic cluster of @xmath9 sites for model i and model ii for the parameters favoring the expected zigzag magnetic order in the ground state . note that spin correlations are systematically reduced , compared to those in a zigzag - ordered phase , by the fact that no symmetry is spontaneously broken in the ground state of a finite cluster .
first , we vary the parameter @xmath40 and change the balance between the fm heisenberg and af kitaev interactions in eqs .
( [ jhk ] ) from @xmath140 to @xmath141 . as the energy increases in this range with increasing @xmath40 @xcite , frustration of spin interactions in eq .
( [ tj ] ) increases as well .
surprisingly , the spectra presented in fig . 6 and in fig .
5(a ) for the selected @xmath114 points in the brillouin zone ( bz ) are only rather weakly influenced by that .
the overall width of somewhat less than @xmath142 does not change showing that it is determined by the hopping term @xmath28 .
the spectral weight is mostly incoherent it concentrates in the central intermediate energy part for the @xmath24 and @xmath25 points at the edge of the bz , while it is moved to high energy for the @xmath104 and @xmath18 points .
a weak low energy quasiparticle ( qp ) is visible at the @xmath24 and @xmath25 points with the spectral weight growing with @xmath40 .
but even at large @xmath141 these qps are rather weak and incoherent spectral weight dominates . the two cases displayed in fig .
6 are nearly identical to each other and to the representative spectrum for @xmath67 presented in fig .
5(a ) . as obtained by ed for @xmath9 sites at selected high - symmetry points @xmath18 , @xmath25 , @xmath24 , and @xmath104 in the bz at @xmath108 and @xmath69 for model ii eq .
( 3 ) , with @xmath143 , and : ( a ) @xmath144 , ( b ) @xmath111 . , width=302 ] second , we consider model ii eq .
( [ j2j3 ] ) and remark that the ed spectra obtained with it are qualitatively similar to those of model i at the same points in the bz , see fig . 7 .
again , the spectral weight is mostly incoherent and moved to high energy at the @xmath18 and @xmath104 points .
here we consider two interaction parameter sets @xmath145 which have less frustration than the case of @xmath111 , @xmath112 and @xmath146 shown in fig .
5(c ) , and the zigzag magnetic order is even more favored . while the spectral functions at the @xmath18 and @xmath104 points are nearly insensitive to these parameter changes , a weak qp forms and is better resolved at the @xmath24 point for both parameter sets of fig .
7 than at @xmath147 .
again , the spectra found at the points belonging to the edge of the bz ( @xmath25 and @xmath24 ) have richer structure than those in model i ( fig . 6 ) . at the @xmath24
point a two - peak structure is found , with a weak qp forming a shoulder of the lower peak at the low energy side .
the broad incoherent spectral weight and its shift to higher energy between the edge and the center of the bz are recognized as universal features for the parameters favoring zigzag magnetic order .
summarizing , the numerical data show that for both models the suppression of low energy qps occurs in a broad range of parameters which correspond to the zigzag magnetic order in the ground state , and is therefore not a result of parameter tuning .
this makes the observations made in the letter robust features of the presented models i and ii . we have found : ( i ) the incoherent distribution of the spectral weight , ( ii ) large spectral weight at high energy , and ( iii ) the absence of low energy qps from the arpes spectra because of destructive interference , even though qp eigenstates exist in the low - energy spectrum .
this could explain pseudogap - type behavior in these materials .
these results suggest that the qualitative agreement with the recently reported experimental angle - resolved photoemission spectroscopy ( arpes ) spectra @xcite is generic , and thus bring new insights into spectral properties of such a strong spin - orbit - coupled mott insulator .
j. chaloupka , g. jackeli , and g. khaliullin , phys .
* 105 * , 027204 ( 2010 ) ; j. reuther , r. thomale , and s. trebst , phys .
b * 84 * , 100406 ( 2011 ) ; r. schaffer , s. bhattacharjee , and y.b .
kim , _ ibid . _ * 86 * , 224417 ( 2012 ) .
r. comin , g. levy , b. ludbrook , z .- h .
zhu , c.n .
veenstra , j.a .
rosen , y. singh , p. gegenwart , d. stricker , j.n .
hancock , d. van der marel , i.s .
elfimov , and a. damascelli , phys .
lett . * 109 * , 266406 ( 2012 ) .
h. gretarsson , j.p .
clancy , x. liu , j.p .
hill , e. bozin , y. singh , s. manni , p. gegenwart , j. kim , a.h .
said , d. casa , t. gog , m.h .
upton , h .- s .
kim , j. yu , v.m .
katukuri , l. hozoi , j. van den brink , and y .- j .
kim , phys .
. lett . * 110 * , 076402 ( 2013 ) .
choi , r. coldea , a.n .
kolmogorov , t. lancaster , i.i .
mazin , s.j .
blundell , p.g .
radaelli , y. singh , p. gegenwart , k.r .
choi , s .- w .
cheong , p.j .
baker , c. stock , and j. taylor , phys .
108 * , 127204 ( 2012 ) .
the zigzag phase is stable in the @xmath148-@xmath16-@xmath17 model on the honeycomb lattice , see : a.f .
albuquerque , d. schwandt , b. hetnyi , s. capponi , m. mambrini , and a.m. luchli , phys .
b * 84 * , 024406 ( 2011 ) ; j. reuther , d.a .
abanin , and r. thomale , _ ibid . _ * 84 * , 014417 ( 2011 ) . g. martinez and p. horsch , phys .
b * 44 * , 317 ( 1991 ) ; j. bona , s. maekawa , and t. tohyama , _ ibid . _
* 76 * , 035121 ( 2007 ) ; i.j .
hamad , a.e .
trumper , a.e .
feiguin , and l.o .
manuel , _ ibid . _ * 77 * , 014410 ( 2008 ) . | we study two heisenberg - kitaev @xmath0-@xmath1-like models on a honeycomb lattice , focusing on the zigzag magnetic phase of na@xmath2iro@xmath3 , and investigate hole motion by exact diagonalization and variational methods .
the spectral functions are quite distinct from those of cuprates and are dominated by large incoherent spectral weight at high energy , almost independent of the microscopic parameters a universal and generic feature for zigzag magnetic correlations .
we explain why quasiparticles at low energy are strongly suppressed in the photoemission spectra and determine an analog of a pseudogap .
we point out that the qualitative features of the predominantly incoherent spectra obtained within the two different models for the zigzag phase are similar , and they have remarkable similarity to recently reported angular resolved photoemission spectra for na@xmath2iro@xmath3 .
frustrated spin systems have long served as a rich source of exotic phenomena such as unconventional magnetic order or spin liquid behavior @xcite .
a beautiful realization of a spin liquid is found in the kitaev model @xcite , where bond - dependent ising interactions lead to strong frustration on the two - dimensional ( 2d ) honeycomb lattice . here antiferromagnetic ( af ) or
ferromagnetic ( fm ) couplings are equivalent ; a fm realization was originally suggested in ref .
@xcite .
the kitaev model is exactly solvable and has spin correlations that vanish beyond nearest neighbor ( nn ) spins @xcite .
this spin liquid is robust against weak time - reversal invariant perturbations , including the heisenberg ones @xcite , similar to the spin - orbital liquid in the su(4 ) symmetric kugel - khomskii model @xcite , but in striking contrast to the 2d compass model @xcite .
such peculiar , highly anisotropic interactions are believed to be realized in @xmath4iro@xmath3 ( @xmath5na , li ) iridates where strong spin - orbit coupling generates kramers doublets of isospin states @xcite .
these systems are mott insulators as confirmed by the electronic structure @xcite .
it has been suggested that effective @xmath6 pseudospins , standing for locally spin - orbital - entangled @xmath7 states @xcite , interact via competing af heisenberg and fm kitaev exchange couplings between nns in the heisenberg - kitaev ( hk ) model this scenario is consistent with the magnetic susceptibility @xcite and with resonant inelastic x - ray scattering ( rixs ) @xcite .
surprisingly , the theoretical predictions of a spin liquid or stripy af phase @xcite were not confirmed but instead zigzag af order , consisting of staggered fm zigzag chains , was found in na@xmath2iro@xmath3 @xcite . to stabilize this phase next - nearest neighbor ( nnn ) and third nearest neighbor ( 3nn ) af heisenberg terms @xcite
have been invoked in the hk model @xcite . recently a simpler scenario , including @xmath8 orbitals in the exchange paths , has been proposed @xcite ; there , the zigzag af phase emerges in a broad range of parameters in the hk model when the signs of all nn exchange terms are _
inverted_. iro@xmath3 : ( a ) cluster of @xmath9 sites , and the elementary translations @xmath10 within two sublattices @xmath11 or @xmath12 ( for two nn @xmath11 atoms @xmath13 ) .
exchange couplings are : in heisenberg ( @xmath14 ) and kitaev ( @xmath15 ) ( solid ) ; heisenberg nnn ( @xmath16 ) and 3nn ( @xmath17 ) ( dashed ) .
( b ) first brillouin zone with high symmetry points : @xmath18 , @xmath19 , @xmath20 ; in exact diagonalization @xmath21 and @xmath22 are equivalent ( @xmath23 ) and called @xmath24 and @xmath25 .
, title="fig:",width=132 ] .5 cm iro@xmath3 : ( a ) cluster of @xmath9 sites , and the elementary translations @xmath10 within two sublattices @xmath11 or @xmath12 ( for two nn @xmath11 atoms @xmath13 ) .
exchange couplings are : in heisenberg ( @xmath14 ) and kitaev ( @xmath15 ) ( solid ) ; heisenberg nnn ( @xmath16 ) and 3nn ( @xmath17 ) ( dashed ) .
( b ) first brillouin zone with high symmetry points : @xmath18 , @xmath19 , @xmath20 ; in exact diagonalization @xmath21 and @xmath22 are equivalent ( @xmath23 ) and called @xmath24 and @xmath25 . , title="fig:",width=139 ]
+ in this letter we investigate whether hole motion in an effective hk model can explain recent angle - resolved photoemission spectroscopy ( arpes ) experiments for na@xmath2iro@xmath3 @xcite .
one expects that a hole might move coherently along the fm zigzag chain [ fig .
1(a ) ] , similar to free hole propagation in states with ferro - order of @xmath7 orbitals @xcite .
it is therefore quite surprising that the arpes spectra for na@xmath2iro@xmath3 are dominated instead by incoherent spectral weight found predominantly at high energy @xcite .
this poses several intriguing questions : ( i ) can one distinguish by arpes the two zigzag states realized by qualitatively different interactions ? ( ii ) is there any similarity between the present arpes results and those known for cuprates @xcite , described by quasiparticles ( qps ) within the @xmath0-@xmath1 model @xcite , or are the spectra dominated by entangled spin - orbital excitations @xcite ? ( iii ) is the incoherent hole scattering on spin excitations as important here as between the fm planes of lamno@xmath3 @xcite ?
we answer them below and show that weak qps exist but are hidden for the present honeycomb lattice so that arpes reveals the incoherent processes .
we consider the following @xmath0-@xmath1-like model ( @xmath26 ) , @xmath27 on the honeycomb lattice [ fig .
[ fig : defs](a ) ] , called _
model i_. the hopping @xmath28 of composite fermions with pseudospin flavor @xmath29 @xcite ( contributions from direct @xmath30-@xmath30 and @xmath30-@xmath31-@xmath30 hopping @xcite ) occurs in the restricted space without double occupancies due to large on - site coulomb repulsion @xmath32 , and the exchange @xmath33 stands for the hk model with fm heisenberg ( @xmath34 ) and af kitaev ( @xmath35 ) exchange , @xmath36 here @xmath1 is the energy unit and @xmath37 $ ] interpolates between the heisenberg and kitaev exchange couplings for nn @xmath38 spins @xmath39 ; the zigzag af phase ( fig .
2 ) is found in a broad range of @xmath40 @xcite .
we also investigate the spectral functions in _ model ii _ , where exchange couplings @xmath41 are extended by nnn and 3nn terms @xmath42 : @xmath43 although our aim is to present the spectral functions obtained by exact diagonalization ( ed ) for a hole in the quantum - fluctuating zigzag af phase , we begin with describing the physical processes and resulting spectral properties of a hole inserted into a fully - polarized ground state @xmath44 , see fig .
2(a ) .
the hole hopping is isotropic , but for convenience we distinguish _ intrachain _ hopping @xmath0 and _ interchain _ hopping @xmath45 .
free hole propagation , see fig .
2(b ) , occurs along the 1d fm zigzag chain when @xmath46 .
it involves two types of ir sites which belong to sublattices @xmath11 and @xmath12 , see fig .
1(a ) .
we are interested in the spectral properties measured in the arpes experiment with the hole creation operator @xmath47 , and we also consider the hole creation on sublattice @xmath11 , @xmath48 @xcite : @xmath49 as we shall see , the sublattice aspect is rather subtle and responsible for the hidden qp states in the arpes experiments for the zigzag phase .
the spectral functions @xmath50 correspond to the physical green s function @xmath51 for @xmath52 , or to the sublattice green s function @xmath53 for @xmath54 ( [ dc ] ) ; here @xmath55 is the energy of the ground state @xmath44 .
iro@xmath3 ( arrows ) .
( a ) the hole ( circle ) may either ( b ) propagate along the fm chain by _ intrachain _ hopping @xmath0 , or ( c ) create a magnon by _ interchain _ hopping @xmath45 .
either this spin defect ( d ) or the hole ( e ) can move along its chain .
( f ) after two steps ( d)+(e ) the hole and the defect may recombine .
broken bonds are indicated by dashed lines .
, width=302 ] below we describe and use a variational approach well adapted to the perturbative regime ( @xmath56 ) to gain more physical insights .
we split the hk @xmath0-@xmath1 model eq .
( 1 ) into an exactly solvable part @xmath57 and the perturbation @xmath58 , @xmath59 .
when exchange interactions are isinglike , either in spin @xcite or in orbital @xcite systems , the number of magnons ( orbitons ) is conserved and the classical ground state @xmath44 is exact .
here we use the same strategy to develop a variational treatment @xcite and include in @xmath57 all terms that conserve the number of magnons , while @xmath58 includes the interchain hole hopping which creates or removes a spin excitation [ fig .
2(c ) ] together with exchange terms generating magnon pairs .
when a hole moves to a neighboring chain , it creates a spin defect with energy @xmath60 , see fig .
2(c ) .
the defect propagates as a magnon [ fig .
2(d ) ] , and the hole can also move by hopping @xmath0 [ fig .
2(e ) ] along the fm chain .
both processes generate two parallel spins on a vertical ( interchain ) bond and increase the magnon energy to @xmath61 , but a hole and a spin defect may also meet at one vertical bond [ fig .
2(f ) ] , which decreases the magnon energy back to @xmath62 . all these processes cause incoherence . other states with several spin defects cost too much energy when @xmath63 these states are neglected in the one - magnon variational treatment . ) in the one - magnon approximation : ( a ) @xmath64 and ( b ) @xmath65 ( [ af ] ) .
the dashed line indicates the static hole energy @xmath66 .
parameters : @xmath67 , @xmath68 and @xmath69 . ,
width=291 ] in the zigzag phase we write the dyson s equation , @xmath70 , for the resolvent , @xmath71 .
the green s functions @xmath72 obtained from it are @xmath73 matrices for two sublattices , @xmath11 and @xmath12 .
the unperturbed green s function , @xmath74 , is found exactly it describes free propagation of a hole along the fm chains , and depends on 1d momentum @xmath75 [ fig .
1(b ) ] , @xmath76 , with @xmath77 being the energy of a static hole . the two @xmath78 states for each @xmath75 result from band folding associated with the two - site unit cell , independent of the 1d nature of hole motion .
these states give two qps at each momentum @xmath75 in @xmath79 , except at @xmath80 where @xmath81 .
in contrast , the lower energy qp , @xmath82 , is _ hidden _ in @xmath83 , while the higher energy one , @xmath84 , has twice larger spectral weight , due to interference between the two sublattice contributions .
the difference between @xmath85 and @xmath86 ( [ af ] ) for model i which follows from states parity is well visible in the weak coupling regime of @xmath63 , see fig .
[ fig : adac1 ] @xcite . at low energy , @xmath87 ,
one finds qps in @xmath64 , but not in @xmath65 .
the width of the qp band is somewhat reduced from the expected @xmath88 , but even stronger renormalization ( its width is lower than @xmath0 ) is found for @xmath89 .
new incoherent features observed in both spectral functions ( [ af ] ) above @xmath90 ( fig . [ fig : adac1 ] ) are generated by a magnon excitation due to the interchain hopping @xmath45 , see fig .
2(c ) .
these states disperse on the scale of @xmath91 ( with energy difference between the respective maxima at @xmath92 and @xmath93 being @xmath94 ) and we recognize here the propagation of a _ holon _ @xcite . ) in the one - magnon approximation : ( a ) @xmath64 , and ( b ) @xmath65 ( [ af ] ) .
the inset shows @xmath75-dependence of the frequencies @xmath95 corresponding to the four distinct features visible in @xmath64 ; the @xmath96 and @xmath97 ones are hidden in @xmath65 .
parameters : @xmath67 , @xmath98 and @xmath69 . ,
width=291 ] in the intermediate coupling regime @xmath99 , the spectral weight moves to higher energies ( fig .
[ fig : adac2 ] ) .
first , one recognizes distinct qps in @xmath85 [ _ hidden _ in @xmath86 ] , with further decreased total width of this subband and the spectral weight decreasing from @xmath92 to @xmath93 .
this follows from hole scattering on magnon excitations ( fig .
2 ) that become gradually more important when @xmath101 increases .
second , the spectra @xmath86 at @xmath98 [ fig .
[ fig : adac2](b ) ] have three notable features : ( i ) the suppressed qp peaks at the onset of the spectrum , ( ii ) the upper holon branch , with spectral weight moving to higher energies when @xmath75 decreases from @xmath93 to @xmath102 , and ( iii ) large spectral weight with weak momentum dependence at the upper edge of the spectrum .
these higher energy features are expected to be strongly renormalized when the constraint to one - magnon excitations is lifted .
in contrast , the qps disappear in the strong coupling regime @xmath103 as well , because the destructive interference is due to parity . indeed , the unbiased ed @xcite performed on a periodic cluster of @xmath9 sites [ fig .
1(a ) ] for model i with the same @xmath67 yields very incoherent spectral weight distribution in @xmath86 , see fig .
[ fig : ed](a ) , in contrast to the 2d @xmath0-@xmath1 model @xcite . as most important feature
, the spectral weight is moved to high energy for the @xmath104 and @xmath18 points . at the @xmath24 point one
observes a broad feature at @xmath105 , accompanied by a shoulder at @xmath106 , and a small qp peak at @xmath107 such a peak is also observed at @xmath25 but absent for momenta inside the brillouin zone ( bz ) .
the spectral weight is transferred from high to lower energy when one moves from the @xmath18 point to the edge of the bz ( @xmath24 and @xmath25 point ) .
these qualitative features are generic and hold in a broad range of parameters ( see the appendix ) . sites at selected high - symmetry points @xmath18 , @xmath25 , @xmath24 , and @xmath104 in the bz at @xmath108 and @xmath69 : ( a ) @xmath109 and ( b ) @xmath110 , both for model i ( 1 ) with @xmath67 ; ( c ) @xmath109 for model ii ( 3 ) , with @xmath111 , @xmath112 and @xmath113 .
arpes spectra for na@xmath2iro@xmath3 @xcite with background subtracted @xcite are shown in ( d ) for the selected @xmath114 points @xcite . ,
title="fig:",width=291 ] .6 cm sites at selected high - symmetry points @xmath18 , @xmath25 , @xmath24 , and @xmath104 in the bz at @xmath108 and @xmath69 : ( a ) @xmath109 and ( b ) @xmath110 , both for model i ( 1 ) with @xmath67 ; ( c ) @xmath109 for model ii ( 3 ) , with @xmath111 , @xmath112 and @xmath113 .
arpes spectra for na@xmath2iro@xmath3 @xcite with background subtracted @xcite are shown in ( d ) for the selected @xmath114 points @xcite . ,
title="fig:",width=291 ] the spectral weight is distributed quite differently in @xmath85 [ fig .
[ fig : ed](b ) ] . here
qp peaks appear for all momenta at energies @xmath115 , with much more weight than those observed in @xmath86 , and there is strong spectral weight transfer to low energies at @xmath104 and @xmath18 points .
the qp weight is maximal for the @xmath18 point and small for @xmath25 and @xmath24 , as expected from the variational approach [ fig .
[ fig : adac2](a ) ] .
we suggest that the lowest energy qp at the @xmath18 point determines the chemical potential @xmath116 and is responsible for the experimentally seen _
pseudogap _ of @xmath117 @xcite .
it is quite remarkable that the ed results obtained with model ii [ fig .
[ fig : ed](c ) ] for parameters suggested @xcite for na@xmath2iro@xmath3 , are qualitatively similar to those found in model i [ fig .
[ fig : ed](a ) ] , but have somewhat richer structure . at the @xmath24 point one finds a two - peak structure with a weak qp at the low energy side , which develops to a shoulder at the @xmath25 point .
the spectra found at @xmath104 and @xmath18 points look similar to those of model i , again with most of the weight concentrated at high energy .
the broad incoherent spectral weight and its shift to higher energy between the @xmath25 and @xmath18 point are recognized here as universal features for the parameters favoring zigzag phase . indeed , moderate changes of parameters ( @xmath40 , and for model ii @xmath118 and @xmath119 ) modifying the degree of frustration while still supporting the zigzag order , do not lead to emerging qps ( see the appendix ) .
we suggest that the ed @xcite simulates here the main features of the arpes spectra measured at @xmath120 k @xcite although the nel temperature @xmath121 k is much smaller due to frustration , the incoherent part of the spectra should be only weakly affected by thermal fluctuations for @xmath122 .
the spectra are incoherent and no qps are seen at the low edge @xmath123 , see fig .
[ fig : ed](d ) . with the total width of the spectrum
@xmath124 we estimate that @xmath125 ev .
large spectral weight at the @xmath18 and @xmath104 points is seen mostly at high energy , as in the ed in @xmath86 but not in @xmath85 [ figs .
[ fig : ed](a ) and [ fig : ed](b ) ] . at @xmath25 and @xmath24 one recognizes three characteristic features with appreciable spectral weight in the arpes data @xcite at @xmath126 @xcite , @xmath127 , and @xmath128 ev which have some correspondence in the ed spectra .
the basic features of the arpes spectra , ( i ) the strong incoherence and ( ii ) the shift of the spectral weight to high energies at the @xmath104 and @xmath18 points , are dictated by the zigzag correlations and are similar in the two models . yet
, we notice differences in the fine - structure which reflect the dramatically different underlying hamiltonians , and therefore suggest that model i is closer to the experimental data @xcite . summarizing
, we have shown that essential features seen in recent arpes experiments for na@xmath2iro@xmath3 @xcite may be described by a universal phenomenology which does not depend on details of modeling .
the framework used is a @xmath0-@xmath1-like model with nearest neighbor kitaev and heisenberg exchange , having conceptually some similarity to high @xmath129 cuprates .
the physical spectral function @xmath86 in the strong coupling regime explains qualitatively the incoherent nature of the arpes spectra , with large spectral weight at high energy and the pseudogap determined by _
hidden _ quasiparticles at low energy .
this also confirms that the low - energy electronic structure of na@xmath2iro@xmath3 can be described by the motion of composite @xmath130 fermions due to strong spin - orbit coupling . _
acknowledgments. _ we thank andrea damascelli , lou - fe feiner , george jackeli , giniyat khaliullin , george a. sawatzky , and ignacio hamad for insightful discussions .
we are grateful to riccardo comin for providing the experimental data of ref .
@xcite .
this work was supported by the max planck ubc centre for quantum materials ; f.t .
thanks advanced materials and process engineering laboratory ( ampel ) , university of british columbia , vancouver for kind hospitality .
a.m.o . kindly acknowledges support by the polish national science center ( ncn ) under project no .
2012/04/a / st3/00331 . |
the desire of higher - density magnetic data storage is becoming more emergent to conform to the explosive growth of today s information industry .
the continued downscaling of storage bits will require significant enhancement of magnetic anisotropy energy ( mae ) per atom to overcome the fundamental , superparamagnetic limit .
single - atom magnetic anisotropy on surfaces serves as an important bottom - up approach of such enhancements .
a surface magnetic adatom no longer has the isotropic orbital electronic structures of its free atom , and consequently its spin - orbit couplings ( soc ) vary in different spin orientations .
such an anisotropy is not only technologically relevant to the smallest magnetic storage bit that can be built , but also of great scientific interest for its fascinating quantum effects @xcite .
pioneering experiments @xcite have been able to measure the mae of single transition - metal atoms and dimers on surfaces , followed by computational studies @xcite that are consistent with the measurements . among those experiments ,
scanning tunneling microscopes ( stm ) can even reach the precision such that magnetic atoms can be built , manipulated , and measured one atom at a time at desired atomic sites of a cun surface @xcite .
there is even a very recent stm study @xcite that can fabricate a bistable atomic - scale antiferromagnet , which enables a low - temperature demonstration of dense nonvolatile storage of information .
however , the magnetic properties of such adatoms , e.g. spin density , have not yet been understood in simple , atomic - scale microscopic pictures .
also , mae of non - transition - metal atoms , the rare earth for example , have not been explored , and are potential candidates with even higher single - atom mae so that one may build atomic - scale storage bits in the future .
we start this study from first - principles calculations of a single rare - earth atom dysprosium ( dy ) on a cun surface .
analysis of subshell quantum numbers , orbital shapes , and occupations of the @xmath2 orbitals leads to simple explanation of the calculated spin - density distribution of the dy atom .
the success of such orbital analysis is then duplicated to the previously studied single fe atom on the same surface @xcite , showing that this simple picture works independently for both the transition - metal @xmath1 and rare - earth @xmath0 orbitals .
this work of dy on a cun surface is also the first study of the mae of a single rare - earth atom on a surface , which is calculated by first principles to be a factor of five larger than the previously largest single - atom mae , co on the pt surface @xcite .
we construct a supercell of 5-layer cu(100 ) slabs , place n and dy atoms on the surface in the same way as the density - functional - theory ( dft ) studies of single fe and mn adatoms @xcite , and then perform dft calculations using the basis set of all - electron full - potential linearized augmented plane wave @xcite . a naive local - density or generalized - gradient approximation makes the @xmath2 orbitals be pinned at the fermi level and causes a fractional @xmath2 occupation , inconsistent with the photoemission measurements of rare - earth nitride bulks , and this problem was shown to be resolved by performing a dft+u calculations instead @xcite .
to find the correct @xmath3 and @xmath4 values for dft+u , we perform pbe+u calculations @xcite and inspect the @xmath2 occupation number @xmath5 of dy on the cun surface with a series of @xmath3 and @xmath4 .
when @xmath6 , and @xmath3 ranges from @xmath7 to @xmath8 ev ( noting that @xmath9 denotes pbe itself ) , the occupation number stays at @xmath10 regardless of the @xmath3 value used .
this means that with solely @xmath3 turned on , the unphysical fermi - level pinning of one of the @xmath0 states remains the same as the naive pbe .
if @xmath4 is turned on to be a positive value @xmath11 ev at @xmath12 ev , the @xmath0 state originally pinned at fermi - level lowers in energy , and becomes fully occupied .
however , the occupation number then becomes @xmath13 , inconsistent with the typical @xmath14 in either a dft+u calculation of a dyn bulk @xcite or a recent quantum - chemistry , complete active space self - consistent field ( casscf ) calculation of a @xmath15 molecule @xcite .
if we first determine @xmath16 ev from a constraint pbe calculation @xcite , and are gradually turning on a negative @xmath4 , we initially find that the dft self - consistent cycles do not converge before @xmath4 reaches @xmath17 ev , indicating unstable electronic structures .
when @xmath4 reaches @xmath17 ev , within a @xmath18@xmath19 cube centered at the dy nucleus , by looking ( left to right ) along @xmath20 , @xmath21 , and @xmath22 directions .
each stick - ball structure corresponds to one of the three directions .
the purple surfaces are the corresponding fe spin density for comparison.,title="fig:",width=264 ] within a @xmath18@xmath19 cube centered at the dy nucleus , by looking ( left to right ) along @xmath20 , @xmath21 , and @xmath22 directions
. each stick - ball structure corresponds to one of the three directions .
the purple surfaces are the corresponding fe spin density for comparison.,title="fig:",width=264 ] the self - consistent cycles converge nicely , and give a reasonable occupation number @xmath23 , consistent with the typical @xmath24 configuration of dy .
therefore , for dy on the cun surface , we have determined that @xmath16 ev and @xmath25 ev yields a @xmath26 configuration that agrees with both the dft+u calculation of a dyn bulk @xcite and the casscf calculation of a @xmath15 molecule @xcite .
atomic charges and spins are calculated by bader analysis @xcite .
as already pointed out in the previous fe study @xcite , when an adatom is deposited onto the cu site of the surface , it establishes polar covalent bonds with the nearest - neighbored n atoms that replaces the original cun binding network .
the calculated electron density of a dy atom in the cun surface is shown in fig .
[ contour]a , together with the previously calculated fe re - presented .
as one can see , the dy atom , sitting even higher on top of the surface , attracts its neighboring n atoms further out of the surface than the fe case .
we have also calculated that dy and its neighboring n are @xmath27 and @xmath28 charged respectively . compared with the @xmath29 and @xmath30 charged fe and its neighboring n respectively , the dy - n bond of the dy system has a polarity approximately the same as the fe - n . by pointing the dy spin in the hollow , n - row , and out - of - plane three symmetry directions ( to be called @xmath20 , @xmath21 , and @xmath22 respectively ) in our pbe+u total - energy calculations with soc included , we obtain the dy mae @xmath31 of @xmath32 , @xmath33 , and @xmath34 , and compare it with the fe case , as shown in the level diagram in fig .
[ contour]b .
one notices that in contrast to the fe case , the most - preferred magnetization axis of dy is oriented in the out - of - plane direction , while both atoms have their least - preferred axis pointing in the hollow direction .
the mae of dy is basically one order of magnitude larger than that of fe , and is five times larger than the @xmath35mev mae of co on the pt surface @xcite , the largest single - atom mae reported previously .
the previously studied fe on the surface has @xmath36 of spin density extends into the surrounding atoms with the spreading primarily along the n - row direction @xcite .
in contrast to fe , when calculating dy on the same surface , we find that a net spin of @xmath37 is localized at the dy atom , and @xmath37 by including the spin of all atoms , indicating that there is no spin spreading .
the @xmath37 of dy includes two parts of contribution from our analysis , @xmath38 from the localized electronic configuration @xmath39 ( within the dy muffin - tin sphere ) , and @xmath40 from the delocalized @xmath41 ( out of the muffin - tin sphere but within the bader basin ) , where @xmath42 denotes a hybridized molecular orbital @xcite .
the significant reduction of spin spreading when replacing fe by dy is obviously because the dy @xmath2 orbitals are more localized than the fe @xmath43 .
another interesting feature is the shape of the spin density . when looking closely at the spin isosurfaces of the dy and fe atoms along all three crystal - symmetry directions in fig .
[ contour]c , the shapes of dy and fe are found to be approximately a hexagon and a square , respectively , along either the n - row or out - of - plane direction , while both atoms become more round - shape - like along the hollow .
this seemly mysterious observation of spin shapes will become clear when we later look into the fe @xmath43 and dy @xmath2 orbitals in the cun surface .
the @xmath1 orbitals in a crystal environment have the following well - known subshell symmetries , @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48
. the exotic , complicated @xmath0 orbitals , which are rarely shown in literatures , have two commonly used sets of subshell symmetries , the cubic and the general sets @xcite . as an example
, we plot the @xmath2 general set in fig .
[ orbital]a .
we find there are four orbitals having a unique six - petal shape : @xmath49 , @xmath50 , @xmath51 , and @xmath52 , and are different from the rest three .
the six - petal shape suggests that the hexagonal spin density is related to the @xmath2 orbitals in some way .
however , the @xmath53 and @xmath54 orbitals that do not have six - petal shapes prevent us from establishing a relation between the spin density and the general - set @xmath2 orbitals . here
we propose an unconventional set of @xmath0 orbitals such that six of the orbitals have fully polarized angular momenta @xmath55 , as one will see its advantages in the orbital analysis of the spin shapes .
five out of the seven orbitals , @xmath56 , @xmath49 , @xmath52 , @xmath51 , and @xmath57 still belong to the general set .
the rest two are @xmath50 and @xmath58 , which are related to the general - set orbitals @xmath53 and @xmath54 simply by the following orthogonal transformation @xmath59 where the primed coordinates are arranged as @xmath60 .
it can be seen that @xmath61 and @xmath62 , which are orthogonally transformed from the general - set @xmath53 and @xmath54 , no longer belong to the general set in the unprimed coordinates , but are actually the general - set orbitals @xmath63 and
@xmath64 in the primed coordinates . one can easily verify that each orbital of this set except for @xmath57 , with its quantization axis @xmath65 properly chosen along one coordinate axis @xmath66 , has its angular momentum fully polarized , that is , they are eigenstates of @xmath67 with eigenvalues @xmath68 . due to the reflection symmetry of these orbitals in the @xmath69 directions ,
their angular - momentum polarization has an sign ambiguity .
the full polarization of @xmath55 of these six @xmath2 orbitals can be further visualized clearly by looking at their orbital shapes . as one can see in fig .
[ orbital]b , the above mentioned six fully - polarized @xmath2 orbitals have an unique six - petal shape , and can be grouped into three pairs .
each pair of orbitals have the same coordinate axis as their central symmetry axis , and are related to each other by exchanging their in - plane axes .
the most obvious full polarization can be realized for the @xmath52 and @xmath51 orbitals , which are eigenstates of @xmath70 with eigenvalues @xmath71 , and have @xmath22 as their central axis .
similarly , each of the rest four fully - polarized orbitals have their central axes as their polarization directions .
the only exception , the @xmath57 orbital , has eight lobes pointing to the corners of a cube .
our calculations show that the dy @xmath2 majority - spin states are all fully occupied and are very low - lying , extremely atomic - like levels .
the orbital analysis of this work does not depend on the details of the dy @xmath2 majority - spin states as long as all of them are fully occupied , and their details are not presented here . for minority - spin states , we plot their pdos in fig .
[ dos]a , and find that @xmath56 and @xmath52 are mainly occupied , and @xmath49 and @xmath51 slightly occupied .
the occupation numbers of the rest of the @xmath2 minority - spin states are negligible .
the @xmath72 electron mentioned in a previous paragraph is mostly delocalized out of the gd muffin - tin sphere , and has a negligible pdos within that sphere .
the dy @xmath2 orbitals are rather localized as indicated from a previous paragraph that they have an approximate @xmath73 radius .
it is a good approximation to think of these orbitals in the way of electronic configurations of an atom , and therefore the nontrivial occupation ( neither fully occupied nor completely empty ) is determined for each orbital from the area under the curve below the fermi level .
the pdos thus implies an approximately @xmath26 configuration for the dy atom itself .
the conventional quantization axes of the fe @xmath43 pdos are oriented in the way that the @xmath20 axis points along the hollow direction , @xmath21 along the n row , and @xmath22 out of plane .
such pdos have been presented in ref .
@xcite , and show that all the majority - spin states are fully occupied .
we re - calculate the pdos and plot the minority spins in fig .
the minority - spin pdos show that @xmath45 is fully occupied , @xmath44 partially occupied , and the rest three minority - spin states basically empty .
the partially occupied spin - minority @xmath44 prohibits us from establishing a simple picture of either the spin - density shape or the soc .
however , one may notice that the @xmath44 and @xmath45 pdos profiles depend nontrivially on the choice ( or interchange ) of the coordinate axes , while the @xmath46 , @xmath47 and @xmath48 ( see the insets of fig . [ dos]b ) have trivially the same set of pdos profiles with interchange of pdos labels corresponding to the interchange of the coordinate axes
. therefore we search for @xmath44 and @xmath45 minority - spin pdos of all three possible assignments of coordinate axes , i.e. , the hollow , n - row , and out - of - plane directions are @xmath20 , @xmath21 , and @xmath22 axes , respectively , and the cyclic permutations . in all three axis assignments ,
all the majority - spin states are always fully occupied , and the @xmath46 , @xmath47 and @xmath48 minority - spin always empty , while the @xmath44 and @xmath45 minority - spin states have specially simple occupations as shown in fig .
[ dos]b under a particular assignment of new axes : the hollow , n - row , and out - of - plane directions are @xmath74 , @xmath75 , and @xmath76 axes , respectively . in this new coordinate system ,
only the @xmath77 ( or equivalently @xmath78 in the old coordinates ) orbital has paired spins , while the rest four all unpaired .
.,title="fig:",width=302 ] .,title="fig:",width=302 ]
with the orbital quantum numbers and occupation numbers determined , we now try to explain the spin - density shape of a dy atom on the cun surface . by starting with the top view ,
one notices that the @xmath51 and @xmath52 orbitals both have the six - petal shapes centered about the @xmath22-axis . as we have identified from the pdos analysis that @xmath52 has roughly paired spins , while @xmath51 has roughly an unpaired spin .
the spin density shape along the @xmath46 plane are therefore dominated by the @xmath51 majority spin alone . observing the shape of the @xmath51 orbital
, one then realizes that the hexagonal shape of the dy spin density from a top view in fig .
[ contour]c is essentially the consequence of the unpaired @xmath51 orbital ( see fig . [
schematic]a ) .
similarly , the @xmath50 and @xmath56 orbitals , both with a six - petal shape centered about the @xmath21-axis , contribute to the hexagonal spin density by looking along the @xmath21 direction . however , the @xmath49 and @xmath58 orbitals , centered about the @xmath20-axis , both have unpaired spins . their two six - petal shapes with @xmath79 relative to each other result in a slightly - round - shape spin - density by looking along the @xmath20 direction .
we can even further understand the unique dent at the @xmath21 ends of the spin isosurface by the following analysis .
the only unpaired orbital that carries a lobe in the @xmath21 direction is @xmath49 , as can be seen in fig .
[ schematic]a , and all the rest have nodes along @xmath21 .
on the other hand , both the other two axes receive lobe contribution from two unpaired - orbitals , e.g. , @xmath51 and @xmath58 both contribute lobes along @xmath20 . consequently , the spin isosurface shrinks in the @xmath21 direction due to relatively less lobe contribution than the @xmath20 and @xmath22 directions . the simple occupation picture of the adatom s localized orbitals that explains the spin - density shape works not only for the rare - earth atom dy .
as we are going to show below , the same picture also works for the previously studied fe adatom . when showing the fe pdos in both the conventional quantization axes ( hollow @xmath20 , n - row @xmath21 , and out - of - plane @xmath22 ) and the unconventional ones ( hollow @xmath74 , n - row @xmath75 , and out - of - plane @xmath76 ) in fig .
[ dos]b , we have demonstrated that in the new coordinate system @xmath80 , only the @xmath77 orbital has paired spins , while the rest four are all unpaired . as seen from fig .
[ schematic]b , the square spin - density shapes of fe in the cun surface from the top view and n - row side view are essentially consequences of the spin - unpaired @xmath81 and @xmath82 orbitals , respectively , while the round shape from the hollow - site side view is the combination of the spin - unpaired @xmath83 and @xmath84 orbitals .
one also notice that the spin isosurface has a dent only at the @xmath85 ends .
similar to the way of explaining the dy spin - isosurface dent , such a dent is the consequence that both the @xmath75 and @xmath76 ends have lobe contributions from the @xmath83 orbital , while @xmath74 has no lobe contribution mainly due to the fact that the @xmath77 orbital is spin - paired .
the stm moving - atom technique has demonstrated its capability of building , manipulating , and measuring a single atomic spin in a well - characterized environment .
first - principles calculations conclude that such an atomic spin forms a surface - embedded molecular magnetic structure @xcite , as well as , reproduce the measured magnetic anisotropy axes @xcite . as an ongoing study of the stm - engineered adatoms ,
this work is the first attempt to understand the spin density of a surface magnetic atom in a simple , atomic - scale microscopic picture .
this is achieved by analyzing the occupations , shapes , and angular momenta of its individual localized orbitals .
these localized orbitals include both the well - known @xmath1 orbitals of the transition - metal atoms and the nontrivial @xmath0 of the rare - earth .
we determine an unconventional set of @xmath2 subshell quantum numbers and an unconventional set of @xmath43 quantization axes that can be used to analyze the shape of the spin density .
the spin - density shape of a magnetic adatom is explained by simply counting the occupation of each individual subshell and spin state of the atom s localized orbital .
these studies provide an important microscopic picture of the atomic - scale origins of magnetization distribution .
we have also done the first computational study of a single rare - earth atom on a surface .
the calculation predicts a record - high mae of @xmath86 mev .
all these theoretical and computational realizations , combined with continuing experimental innovations @xcite , may lead to further engineering of the single - atom magnetic anisotropy , further exploration of the smallest magnetic storage bit , and new fascinating applications of atomic - scale magnetism .
we thank a. heinrich , b. jones , and d. rugar for stimulating discussions .
we appreciate valuable technical support from j. l. li .
we acknowledge financial supports from the taiwan national science foundation , the taiwan ministry of education , and the taiwan national center for theoretical sciences .
we also acknowledge the support of computing facilities provided by the taiwan national center for high - performance computing , 99 j. r. friedman , m. p. sarachik , j. tejada and r. ziolo , phys .
lett . * 76 * , 3830 ( 1996 ) .
l. thomas , f. lionti , r. ballou , d. gatteschi , r. sessoli and b. barbara , nature * 383 * , 145 ( 1996 ) . w. wernsdorfer and r. sessoli , science * 284 * , 133 ( 1999 ) .
p. gambardella _ et al .
_ , science * 300 * , 1130 ( 2003 ) .
c. f. hirjibehedin _ et al .
_ , science * 317 * , 1199,(2007 ) .
a. f. otte _ et al .
_ , nature physics , * 4 * , 847,(2008 ) .
loth , s. ; etzkorn , m. ; lutz , c. p. ; eigler , d. m. ; heinrich , a. j. _ science _ * 2010 * , 329 , 1628 , ( 2010 )
. b. lazarovits , l. szunyogh and p. weinberger , phys .
b * 67 * , 024415 , ( 2003 ) .
a. b. shick , f. m@xmath87ca and a. i. lichtenstein , phys .
b * 79 * , 172409 , ( 2009 ) .
s. loth , s. baumann , c. p. lutz , d. m. eigler , a. j. heinrich , science * 335 * , 196 ( 2012 ) .
r. @xmath88itko and t. pruschke , new j. phys .
* 12 * , 063040 , ( 2010 ) . c. y. lin and b. a. jones , phys .
b * 83 * , 014413 , ( 2011 ) .
p. blaha , k. schwarz , g. madsen , d. kvasnicka , and j. luitz , computer code wien2k : an augmented plane wave plus local orbitals program for calculating crystal properties .
( vienna university of technology , 2001 ) .
p. larson , w. r. l. lambrecht , a. chantis and m. schilfgaarde , phys .
b * 75 * , 045114 , ( 2007 ) .
anisimov , v. i. ; solovyev , i. v. ; korotin , m. a. ; czyzyk , m. t. ; sawatzky , g. a. _ phys .
b _ * 1993*,48 , 16929 , ( 1993 ) .
a. i. liechtenstein , v. i. anisimov and j. zaanen , phys . rev .
b * 52 * , r5467 , ( 1995 ) .
j. p. perdew , k. burke and m. ernzerhof , phys .
* 77 * , 3865 , ( 1996 ) .
g. lanza , z. varga , m. kolonits and m. hargittaia , j. chem . phys .
* 128 * , 074301 , ( 2008 ) .
g. k. h. madsen and p. nov@xmath87k , europhys . lett . *
69 * , 777 , ( 2005 ) .
r. f. w. bader _ atoms in molecules a quantum theory _ , ( clarendon press , oxford , 1990 ) .
the dy @xmath2 , @xmath89 , and @xmath90 have radii of @xmath73 , @xmath91 , and @xmath92 , respectively . as a result
, the @xmath2 contributes a much larger radial part @xmath93 to the soc than @xmath42 .
in fact , the @xmath42 orbital is mainly distributed out of the dy muffin - tin sphere , and its soc is not included in this calculation .
see , for example , http://winter.group.shef.ac.uk/orbitron | studying single - atom magnetic anisotropy on surfaces enables the exploration of the smallest magnetic storage bit that can be built . in this work , magnetic anisotropy of a single rare - earth atom on a surface
is studied computationally for the first time .
the single adatom and its substrate surface are chosen to be a dysprosium ( dy ) atom and a copper - nitrite surface , respectively , where single transition - metal magnetic atoms on the same surface were previously studied one atom at a time by scanning tunneling microscopes .
we propose unconventional @xmath0 and @xmath1 subshell symmetries so that following the first - principles calculations , simple pictorial analyses of the spin - density distribution can be performed for the first time , independently for both a rare - earth atom dy and a transition - metal fe .
the magnetic anisotropy energy of dy on the surface is calculated to be a factor of five larger than the previous highest one , reaching a record - high value of 31 mev . |
let @xmath2 denote a space of constant curvature , either the @xmath3-dimensional sphere @xmath4 , euclidean space @xmath5 , or hyperbolic space @xmath6 with @xmath7 . in discrete geometry
, it is commonly asked to find the highest possible packing density in @xmath2 by congruent non - overlapping balls of a given radius @xcite , @xcite .
euclidean cases are the best explored .
for example , the densest possible lattice packings are known for @xmath8 through @xmath9 . in higher dimensions , however , mostly only bounds are known .
furthermore , no sharp bounds exist for irregular packings in @xmath5 when @xmath10 .
one major recent development has been the settling of the long - standing kepler conjecture , part of hilbert s 18th problem , by thomas hales at the turn of the 21st century .
computer - assisted proof was largely based on a program set forth by l. fejes tth in the 1950 s @xcite .
the definition of packing density is critical in hyperbolic space as shown by brczky @xcite . for other standard examples
see also @xcite , @xcite . the most widely accepted notion of packing density considers the local densities of balls with respect to their dirichlet voronoi cells ( cf .
@xcite and @xcite ) .
in order to consider horoball packings in @xmath11 , we use an extended notion of such local density .
let @xmath12 be a horoball in packing @xmath13 , and @xmath14 be an arbitrary point .
define @xmath15 to be the perpendicular distance from point @xmath16 to the horosphere @xmath17 , where @xmath15 is taken to be negative when @xmath18 .
the dirichlet voronoi cell @xmath19 of a horoball @xmath12 is defined as the convex body @xmath20 both @xmath12 and @xmath21 are of infinite volume , so the usual notion of local density is modified as follows .
let @xmath22 denote the ideal center of @xmath12 at infinity , and take its boundary @xmath23 to be the one - point compactification of euclidean @xmath24-space .
let @xmath25 be the euclidean @xmath26-ball with center @xmath27 .
then @xmath28 and @xmath29 determine a convex cone @xmath30 with apex @xmath31 consisting of all hyperbolic geodesics passing through @xmath29 with limit point @xmath31 .
the local density @xmath32 of @xmath12 to @xmath21 is defined as @xmath33 this limit is independent of the choice of center @xmath34 for @xmath35 . in the case of periodic ball or horoball packings , the local density defined above
can be extended to the entire hyperbolic space .
this local density is related to the simplicial density function ( defined below ) that we generalized in @xcite and @xcite . in this paper
we will use such definition of packing density ( cf .
section 3 ) . the alternate method suggested by bowen and radin @xcite , @xcite uses nevo s point - wise ergodic theorem to assure that the standard euclidean limit notion of density is well - defined for @xmath6 .
first they define a metric on the space @xmath36 of relatively - dense packings by compact objects , based on hausdorff distance , corresponding to uniform convergence on compact subsets of @xmath6 .
then they study the measures invariant under isometries of @xmath36 rather than individual packings .
there is a large class of packings of compact objects in hyperbolic space for which such density is well - defined . using ergodic methods ,
they show that if there is only one optimally dense packing of @xmath5 or @xmath6 , up to congruence , by congruent copies of bodies from some fixed finite collection , then that packing must have a symmetry group with compact fundamental domain .
moreover , for almost any radius @xmath37 the optimal ball packing in @xmath6 has low symmetry .
a coxeter simplex is an @xmath3-dimensional simplex in @xmath2 such that its dihedral angles are either submultiples of @xmath38 , or zero .
the group generated by reflections on the sides of a coxeter simplex is called a coxeter simplex reflection group .
such reflections give a discrete group of isometries of @xmath2 with the coxeter simplex as its fundamental domain ; hence the groups give regular tessellations of @xmath2 .
the coxeter groups are finite for @xmath4 , and infinite for @xmath5 or @xmath6 . in @xmath6
we allow unbounded simplices with ideal vertices at infinity @xmath39 .
coxeter simplices exist only for dimensions @xmath40 ; furthermore , only a finite number exist in dimensions @xmath41 .
_ @xcite computed the volumes of all coxeter simplices in hyperbolic @xmath3-space , see also kellerhals @xcite .
such simplices are the most elementary building blocks of hyperbolic manifolds , the volume of which is an important topological invariant . in the @xmath3-dimensional space @xmath2 of constant curvature @xmath42 ,
define the simplicial density function @xmath43 to be the density of @xmath44 spheres of radius @xmath45 mutually touching one another with respect to the simplex spanned by the centers of the spheres .
l. fejes tth and h. s. m. coxeter conjectured that the packing density of balls of radius @xmath45 in @xmath2 can not exceed @xmath43 .
rogers @xcite proved this conjecture in euclidean space @xmath5 .
the @xmath46-dimensional spherical case was settled by l. fejes tth @xcite , and brczky @xcite , who proved the following extension : in an @xmath3-dimensional space of constant curvature , consider a packing of spheres of radius @xmath45 .
in the case of spherical space , assume that @xmath47 .
then the density of each sphere in its dirichlet voronoi cell can not exceed the density of @xmath44 spheres of radius @xmath45 mutually touching one another with respect to the simplex spanned by their centers .
in hyperbolic space , the monotonicity of @xmath48 was proved by brczky and florian @xcite ; marshall @xcite showed that for sufficiently large @xmath3 , function @xmath43 is strictly increasing in variable @xmath45 .
kellerhals @xcite showed @xmath49 , and that in cases considered by marshall the local density of each ball in its dirichlet voronoi cell is bounded above by the simplicial horoball density @xmath50 .
this upper bound for density in hyperbolic space @xmath51 is @xmath52 , which is not realized by packing regular balls .
however , it is attained by a horoball packing of @xmath53 where the ideal centers of horoballs lie on an absolute figure of @xmath53 ; for example , they may lie at the vertices of the ideal regular simplex tiling with coxeter - schlfli symbol @xmath54 .
in @xcite we proved that the optimal ball packing arrangement in @xmath51 mentioned above is not unique .
we gave several new examples of horoball packing arrangements based on totally asymptotic coxeter tilings that yield the brczky
florian upper bound @xcite .
furthermore , in @xcite , @xcite we found that by allowing horoballs of different types at each vertex of a totally asymptotic simplex and generalizing the simplicial density function to @xmath6 for @xmath55 , the brczky - type density upper bound is no longer valid for the fully asymptotic simplices for @xmath56 .
for example , in @xmath57 the locally optimal packing density is @xmath58 , higher than the brczky - type density upper bound of @xmath59 .
however these ball packing configurations are only locally optimal and can not be extended to the entirety of the hyperbolic spaces @xmath6 .
further open problems and conjectures on @xmath0-dimensional hyperbolic packings are discussed in @xcite . a recent result of jacquemet @xcite
gives a formula for the inradius of a hyperbolic truncated @xmath3-simplex based on its gram matrix .
the second - named author has several additional results on globally and locally optimal ball packings in @xmath6 , @xmath4 , and the eight thurston geomerties arising from thurston s geometrization conjecture @xcite , @xcite , @xcite , @xcite , @xcite , @xcite .
these packing densities are global or local , depending on whether the density obtained in a fundamental domain can or can not be extended to the entire space . in this paper
we continue our investigations on ball packings in hyperbolic 4-space . using horoball packings ,
allowing horoballs of different types , we find seven counterexamples ( realized by allowing up to three horoball types ) to one of l. fejes tth s conjectures stated in the concluding section of his book regular figures : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ finally we draw attention to the tessellations @xmath60 of 4-dimensional hyperbolic space , the cell - inspheres and cell - circumspheres of which are also expected to form a closest packing and loosest covering .
the corresponding densities are @xmath61 and @xmath62 @xcite _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
in this paper we use the cayley klein ball model , and a projective interpretation of hyperbolic geometry .
this has the advantage of greatly simplifying our calculations in higher dimensions as compared to other models such as the poincar model . in this section
we give a brief review of the concepts used in this paper . for a general discussion and background in hyperbolic geometry and
the projective models of thurston geometries see @xcite and @xcite .
we use the projective model in lorentzian @xmath63-space @xmath64 of signature @xmath65 , i.e. @xmath64 is the real vector space @xmath66 equipped with the bilinear form of signature @xmath65 @xmath67 where the non - zero real vectors @xmath68 and @xmath69 @xmath70 represent points in projective space @xmath71 .
@xmath6 is represented as the interior of the absolute quadratic form @xmath72\in\mathcal{p}^n | \langle ~ \mathbf{x},~\mathbf{x } \rangle = 0 \}=\partial \mathbb{h}^n\ ] ] in real projective space @xmath73 .
all proper interior points @xmath74 satisfy @xmath75 .
the boundary points @xmath76 in @xmath77 represent the absolute points at infinity of @xmath6 .
points @xmath78 satisfying @xmath79 lie outside @xmath76 and are referred to as outer points of @xmath6 . take @xmath80 ) \in \mathcal{p}^n$ ] , point @xmath81 \in \mathcal{p}^n$ ] is said to be conjugate to @xmath82 $ ] relative to @xmath31 when @xmath83 .
the set of all points conjugate to @xmath80)$ ] form a projective ( polar ) hyperplane @xmath84\in\mathcal{p}^n | \langle ~ \mathbf{x},~\mathbf{y } \rangle = 0 \}.\ ] ] hence the bilinear form @xmath31 in ( [ bilinear_form ] ) induces a bijection or linear polarity @xmath85 between the points of @xmath77 and its hyperplanes .
point @xmath86 $ ] and hyperplane @xmath87 $ ] are incident if the value of the linear form @xmath88 evaluated on vector @xmath89 is zero , i.e. @xmath90 where @xmath91 , and @xmath92 .
similarly , lines in @xmath77 are characterized by 2-subspaces of @xmath66 or @xmath26-spaces of @xmath93 @xcite .
let @xmath94 denote a polyhedron bounded by a finite set of hyperplanes @xmath95 with unit normal vectors @xmath96 directed towards the interior of @xmath16 : @xmath97 in this paper @xmath16 is assumed to be an acute - angled polyhedron with proper or ideal vertices .
the grammian matrix @xmath98 is an indecomposable symmetric matrix of signature @xmath65 with entries @xmath99 and @xmath100 for @xmath101 where @xmath102 this is visualized using the weighted graph or scheme of the polytope @xmath103 .
the graph nodes correspond to the hyperplanes @xmath95 and are connected if @xmath95 and @xmath104 not perpendicular ( @xmath105 ) .
if they are connected we write the positive weight @xmath106 where @xmath107 on the edge , and unlabeled edges denote an angle of @xmath108 . for examples , see the coxeter diagrams in table [ table : simplex_list ] .
in this paper we set the sectional curvature of @xmath6 , @xmath109 , to be @xmath110 .
the distance @xmath111 of two proper points @xmath82 $ ] and @xmath81 $ ] is calculated by the formula @xmath112 the perpendicular foot @xmath113 $ ] of point @xmath114 $ ] dropped onto plane @xmath115 $ ] is given by @xmath116 where @xmath117 is the pole of the plane @xmath115 $ ] . a horosphere in @xmath6 ( @xmath118 is a hyperbolic @xmath3-sphere with infinite radius centered at an ideal point on @xmath39 .
equivalently , a horosphere is an @xmath26-surface orthogonal to the set of parallel straight lines passing through a point of the absolute quadratic surface .
a horoball is a horosphere together with its interior . in order to derive the equation of a horosphere
, we introduce a projective coordinate system for @xmath77 with a vector basis @xmath119 so that the cayley - klein ball model of @xmath6 is centered at @xmath120 , and set an arbitrary point at infinity to lie at @xmath121 .
the equation of a horosphere with center @xmath122 passing through point @xmath123 is derived from the equation of the the absolute sphere @xmath124 , and the plane @xmath125 tangent to the absolute sphere at @xmath122 .
the general equation of the horosphere is @xmath126 plugging in for @xmath23 we obtain @xmath127 if @xmath128 , the equation of a horosphere in projective coordinates is @xmath129 and in cartesian coordinates setting @xmath130 it becomes @xmath131 in an @xmath3-dimensional hyperbolic space any two horoballs are congruent in the classical sense : each have an infinite radius .
however , it is often useful to distinguish between certain horoballs of a packing .
we use the notion of horoball type with respect to the packing as introduced in @xcite .
two horoballs of a horoball packing are said to be of the _ same type _ or _ equipacked _ if and only if their local packing densities with respect to a given cell ( in our case a coxeter simplex ) are equal .
if this is not the case , then we say the two horoballs are of _ different type_. for example , in the above discussion horoballs centered at @xmath122 passing through @xmath23 with different values for the final coordinate @xmath132 are of different type relative to an appropriate cell . in order to compute volumes of horoball pieces , we use jnos bolyai s classical formulas from the mid 19-th century : 1 . the hyperbolic length @xmath133 of a horospheric arc that belongs to a chord segment of length @xmath134 is @xmath135 2 .
the intrinsic geometry of a horosphere is euclidean , so the @xmath26-dimensional volume @xmath136 of a polyhedron @xmath137 on the surface of the horosphere can be calculated as in @xmath138 .
the volume of the horoball piece @xmath139 determined by @xmath137 and the aggregate of axes drawn from @xmath137 to the center of the horoball is @xmath140
let @xmath141 be a coxeter tiling . a rigid motion mapping one cell of @xmath141 onto another maps the entire tiling onto itself .
the symmetry group of a coxeter tiling is its coxeter group , denoted by @xmath142 .
any simplex cell of @xmath141 acts as a fundamental domain @xmath143 of @xmath142 , where the coxeter group is generated by reflections on the @xmath24-dimensional facets of @xmath143 . in this paper
we consider only asymptotic coxter simplices , i.e. ones that have at least one ideal vertex . in table
[ table : simplex_list ] we list the nine asymptotic coxeter simplices that exist in hyperbolic 4-space , together with their volumes . for a complete discussion of hyperbolic coxeter simplices and their volumes for dimensions @xmath41 , see johnson _
_ @xcite .
we define the density of a horoball packing @xmath144 of a coxeter simplex tiling @xmath141 as @xmath145 here @xmath143 denotes the simplicial fundamental domain of tiling @xmath141 , @xmath3 is the number of ideal vertices of @xmath146 , and @xmath147 are the horoballs centered at ideal vertices .
we allow horoballs of different types at the asymptotic vertices of the tiling .
a horoball type is allowed if it yields a packing : no two horoballs may have an interior point in common .
in addition we require that no horoball extend beyond the facet opposite the vertex where it is centered so that the packing remains invariant under the actions of the coxeter group of the tiling .
if these conditions are satisfied , we can extend the packing density from the simplicial fundamental domain @xmath143 to the entire @xmath148 using the coxeter group @xmath141 associated with a tiling . in the case of coxeter
simplex tilings , dirichlet voronoi cells coincide with the coxeter simplices .
we denote the optimal horoball packing density as @xmath149 diagram & notation & symbol & volume & density + simply asymptotic & & & & + ( 0,0 ) ( 1,0 ) ; ( 1,0 ) ( 1.5,0.25 ) ; ( 1,0 ) ( 1.5,-0.25 ) ; ( 0,0 ) circle ( .05 ) ; ( 0.5,0 ) circle ( .05 ) ; ( 1,0 ) circle ( .05 ) ; ( 1.5,0.25 ) circle ( .05 ) ; ( 1.5,-0.25 ) circle ( .05 ) ; at ( 1,-0.175 ) @xmath0 ; & @xmath150 $ ] & @xmath151 & @xmath152 & 0.71644896 + ( 0,0 ) ( .5,0 ) ; ( .5,0 ) ( 1,0.25 ) ; ( .5,0 ) ( 1,-0.25 ) ; ( 1,0.25 ) ( 1.5,0 ) ; ( 1,-0.25 ) ( 1.5,0 ) ; ( 0,0 ) circle(.05 ) ; ( .5,0 ) circle(.05 ) ; ( 1,0.25 ) circle(.05 ) ; ( 1,-0.25 ) circle(.05 ) ; ( 1.5,0 ) circle(.05 ) ; & @xmath153}]$ ] & @xmath154 & @xmath155 & 0.71644896 + ( 0,0 )
( 2,0 ) ; ( 0,0 ) circle(.05 ) ; ( 0.5,0 ) circle(.05 ) ; ( 1,0 ) circle(.05 ) ; ( 1.5,0 ) circle(.05 ) ; ( 2,0 ) circle(.05 ) ; at ( .75,0.15 ) @xmath0 ; at ( 1.75,0.15 ) @xmath0 ; & @xmath156 $ ] & @xmath157 & @xmath158 & 0.60792710 + ( 0,0 )
( 1,0 ) ; ( 1,0 )
( 1.5,0.25 ) ; ( 1,0 )
( 1.5,-0.25 ) ; ( 0,0 ) circle ( .05 ) ; ( .5,0 ) circle ( .05 ) ; ( 1,0 ) circle ( .05 ) ; ( 1.5,0.25 ) circle ( .05 ) ; ( 1.5,-0.25 ) circle ( .05 ) ; at ( .75,0.175 ) @xmath0 ; & @xmath159 $ ] & @xmath160 & @xmath161 & 0.60792710 + ( 0,0 ) ( 1,0 ) ; ( 0,0 ) ( 0.25,.5 ) ; ( 0.25,.5 ) ( .75,.5 ) ; ( .75,.5 ) ( 1,0 ) ; ( 0,0 ) circle(.05 ) ; ( .5,0 ) circle(.05 ) ; ( 1,0 ) circle(.05 ) ; ( 0.25,.5 ) circle(.05 ) ; ( .75,.5 ) circle(.05 ) ; at ( 0,0.275 ) @xmath0 ; at ( 1,0.275 ) @xmath0 ; & @xmath162 $ ] & @xmath163 & @xmath164 & 0.71644896 + doubly asymptotic & & & & + ( 0,0 ) ( 1,0 ) ; ( 1,0 ) ( 1.5,0.25 ) ; ( 1,0 ) ( 1.5,-0.25 ) ; ( 0,0 ) circle ( .05 ) ; ( .5,0 ) circle ( .05 ) ; ( 1,0 ) circle ( .05 ) ; ( 1.5,0.25 ) circle ( .05 ) ; ( 1.5,-0.25 ) circle ( .05 ) ; at ( 1,-0.2 ) @xmath0 ; at ( 0.25,0.15 ) @xmath0 ; & @xmath165 $ ] & @xmath166 & @xmath167 & 0.71644896 + ( 0,0 )
( .5,0 ) ; ( .5,0 ) ( 1,0.25 ) ; ( .5,0 ) ( 1,-0.25 ) ; ( 1,0.25 ) ( 1.5,0 ) ; ( 1,-0.25 )
( 1.5,0 ) ; ( 0,0 ) circle(.05 ) ; ( .5,0 ) circle(.05 ) ; ( 1,0.25 ) circle(.05 ) ; ( 1,-0.25 ) circle(.05 ) ; ( 1.5,0 ) circle(.05 ) ; at ( 0.25,0.15 ) @xmath0 ; & @xmath168}]$ ] & @xmath169 & @xmath170 & 0.71644896 + triply asymptotic & & & & + ( 0,0.25 )
( .5,0 ) ; ( 0,-0.25 ) ( .5,0 ) ; ( .5,0 )
( 1 , 0.25 ) ; ( .5,0 ) ( 1,-0.25 ) ; ( 0,0.25 ) circle ( .05 ) ; ( 0,-0.25 ) circle ( .05 ) ; ( .5,0 ) circle ( .05 ) ; ( 1,0.25 ) circle ( .05 ) ; ( 1,-0.25 ) circle ( .05 ) ; at ( .75,-0.3 ) @xmath0 ; & @xmath171 $ ] & @xmath172 & @xmath170 & 0.71644896 + ( 0,0 ) ( .5,0.25 ) ; ( 0,0 ) ( .5,-0.25 ) ; ( .5,0.25 ) ( .5,0 ) ; ( .5,0.25 ) ( 1,0 ) ; ( .5,-0.25 ) ( .5,0 ) ; ( .5,-0.25 ) ( 1,0 ) ; ( 0,0 ) circle(.05 ) ; ( .5,0 ) circle(.05 ) ; ( 1,0 ) circle(.05 ) ; ( .5,0.25 ) circle(.05 ) ; ( .5,-0.25 ) circle(.05 ) ; & @xmath173\times[]}]$ ] & @xmath174 & @xmath175 & 0.71644896 + .
the number of stars in the superscript @xmath176 and @xmath177 indicates that the fundamental simplex of the group has two or three ideal vertices.,height=226 ] the asymptotic coxeter simplex tilings are related through the subgroup structure of their coxeter symmetry groups as shown in figure [ fig : lattice_of_subgroups ] @xcite , @xcite .
let @xmath178 and @xmath179 be the two coxeter symmetry groups of coxeter tilings @xmath180 and @xmath181 , respectively . when the index of coxeter group @xmath178 in @xmath179 is two , i.e. @xmath182 , then the two coxeter groups differ by one reflection , and the fundamental domain of @xmath179 is obtained from that of @xmath178 by domain doubling , that is by merging a certain pair of neighboring domains by removing a common facet . in the case of asymptotic coxeter simplices if @xmath182 and the number of asymptotic vertices of the fundamental domains of @xmath178 and @xmath179 are equal , then the new fundamental domain is obtained by removing a facet adjacent to an asymptotic vertex .
if the number of asymptotic vertices increases by one , then the cells of @xmath179 are obtained by removing a facet opposite to the asymptotic vertices of @xmath178 and merging the cells .
the relationship between the volumes of the cells of @xmath178 and @xmath179 when @xmath183 is given by @xmath184 .
if the index of the groups is two , then a packing density @xmath185 for the bigger group can be extended to the smaller group @xmath179 .
we compute the optimal horoball packing density for the coxeter simplex tiling @xmath151 ; the other simply asymptotic cases can be obtained using the same method .
case @xmath157 was computed by the second - named author in @xcite .
[ proposition : s4 ] the optimal horoball packing density for simply asymptotic coxeter simplex tiling @xmath186 is @xmath187 .
let @xmath188 be the simplicial fundamental domain of coxeter tiling @xmath189 .
we set coordinates for its vertices @xmath190 that satisfy the angle requirements .
our choice of vertices , as well as forms for hyperplanes @xmath191 $ ] opposite to vertices @xmath192 , are given in table [ table : data_s_a ] . in order to maximize the packing density ,
we determine the largest horoball type @xmath193 centered at ideal vertex @xmath122 that is admissible in cell @xmath194 .
this is the horoball with type - parameter @xmath132 ( intuitively the radius " of the horoball ) such that the horoball @xmath193 is tangent to the plane of the hyperface @xmath195 $ ] bounding the fundamental simplex opposite of @xmath122 .
the perpendicular foot @xmath196 $ ] of vertex @xmath122 on plane @xmath195 $ ] , @xmath197 is the point of tangency of horoball @xmath193 and hyperface @xmath198 of the the simplex cell .
plugging in for @xmath199 and solving equation ( [ eqn : horosphere1 ] ) , we find that the horoball with type - parameter @xmath200 is the optimal type .
the equation of horosphere @xmath201 centered at @xmath122 passing through @xmath199 is @xmath202 intersecting the sides of the simplex at @xmath203 , @xmath204 , and @xmath205 .
( b ) horospheric tetrahedron on hyperface opposite @xmath122.,title="fig:",height=188 ] ( a ) ( b ) the intersections @xmath206 $ ] of horosphere @xmath207 and simplex edges are found by parameterizing the simplex edges as @xmath208 @xmath209 , and computing their intersections with @xmath207 .
see figure ( [ fig : horospheric_triangle ] ) , and table [ table : data_s_a ] for the intersection points .
the volume of the horospherical tetrahedron determines the volume of the horoball piece by equation . in order to determine the data of the horospheric tetrahedron , we compute the hyperbolic distances @xmath210 by the formula ( 5 ) @xmath211 where @xmath212 .
moreover , the horospherical distances @xmath213 can be calculated by formula .
the intrinsic geometry of the horosphere is euclidean , so we use the cayley - menger determinant to find the volume @xmath136 of the horospheric tetrahedron @xmath137 , @xmath214 the volume of the optimal horoball piece in the fundamental simplex is @xmath215 hence by the coxeter group @xmath216 the optimal horoball packing density of the coxeter simplex tiling @xmath189 becomes @xmath217 the same method is used to find the optimal packing density of the remaining simply asymptotic coxeter simplex tilings .
results of the computations are given in table [ table : data_s_a ] .
we summarize the results : the optimal horoball packing density for simply asymptotic coxeter simplex tiling @xmath218 , @xmath219 is @xmath220 .
witt symb . &
@xmath151 & @xmath154 & @xmath157 & @xmath160 & @xmath163 + + @xmath122 & @xmath221 & @xmath221 & @xmath221 & @xmath221 & @xmath221 + @xmath222 & @xmath223 & @xmath224&@xmath225&@xmath226&@xmath227 + @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath232 & @xmath233 + @xmath234 & @xmath235 & @xmath236 & @xmath237 & @xmath238 & @xmath229 + @xmath239 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath244 + + @xmath245 & @xmath246 & @xmath247 & @xmath248 & @xmath249 & @xmath250 + @xmath251 & @xmath252 & @xmath253 & @xmath254 & @xmath255 & @xmath256 + @xmath257 & @xmath258 & @xmath259 & @xmath260 & @xmath261 & @xmath262 + @xmath263 & @xmath264 & @xmath265 & @xmath266 & @xmath267 & @xmath268 + @xmath269 & @xmath270 & @xmath270 & @xmath271 & @xmath272 & @xmath270 + + @xmath132 & @xmath273 & @xmath274 & @xmath275 & @xmath276 & @xmath277 + + @xmath203 & @xmath278 & @xmath279 & @xmath280 & @xmath281 & @xmath282 + @xmath204 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & @xmath287 + @xmath205 & @xmath288 & @xmath289 & @xmath290 & @xmath291 & @xmath292 + @xmath293 & @xmath294 & @xmath295 & @xmath242 & @xmath296 & @xmath297 + + @xmath298 & @xmath299 & @xmath300 & @xmath301 & @xmath302 & @xmath303 + + @xmath304 & @xmath305 & @xmath305 & @xmath306 & @xmath306 & @xmath305 + in cases where the coxeter simplex has multiple asymptotic vertices , we allow horoballs of different types at each vertex .
the equations of horospheres centered at @xmath307 @xmath308 and @xmath309 where @xmath130 are @xmath310 and @xmath311 as in simply asymptotic cases , we first find bounds for the largest possible horoball type admissible at each asymptotic vertex . such a horoball is tangent to the facet opposite its center .
next we set one horoball to be of the largest type , and increase the size of the other horoballs until they become tangent or inadmissible .
we then vary the types of the horoballs within the allowable range to find the optimal packing density .
the following lemma was proved in @xcite and it gives the relationship between the volumes of two tangent horoball pieces centered at certain vertices of a tiling , as we continuously vary their types .
let @xmath312 and @xmath313 be two congruent @xmath3-dimensional convex pyramid - like regions with vertices at @xmath314 and @xmath315 that share the common edge @xmath316 .
let @xmath317 and @xmath318 denote two horoballs centered at @xmath314 and @xmath315 tangent at point @xmath319 .
define the point of tangency @xmath320 ( the midpoint " ) such that the equality @xmath321 holds for the volumes of the horoball sectors .
see figure [ fig:2horoballs ] ( a ) .
[ lemma : szirmai ] let @xmath134 be the hyperbolic distance between @xmath320 and @xmath322 , then @xmath323 strictly increases as @xmath324 . and @xmath315 tangent at point @xmath325 .
( b ) doubly asymptotic case @xmath326 , the horoball types of @xmath327 and @xmath328 can be varied between the two limiting cases .
the case where @xmath328 is maximal is pictured.,title="fig:",height=75 ] and @xmath315 tangent at point @xmath325 .
( b ) doubly asymptotic case @xmath326 , the horoball types of @xmath327 and @xmath328 can be varied between the two limiting cases .
the case where @xmath328 is maximal is pictured.,title="fig:",height=151 ] ( a ) ( b ) the optimal horoball packing density for coxeter simplex tiling @xmath329 is @xmath330 .
parameterize the fundamental domain @xmath331 according to table [ table : data_mult ] , so that the two asymptotic vertices are at two opposite poles of the ball model .
let the corner of the simplex at @xmath122 be @xmath332 and that at @xmath234 be @xmath333 .
place two horoballs @xmath334 and @xmath335 that pass through @xmath336 and @xmath337 respectively at @xmath122 and @xmath234 .
let @xmath338 denote the hyperbolic distance of the center of the model @xmath242 and the point @xmath339 .
if the two horoballs are of maximal type , @xmath340 and @xmath341 are tangent to their respective hyperfaces @xmath342 $ ] and @xmath343 $ ] , then their interiors intersect so the packing density is optimal when the horoballs are tangent at one point .
set @xmath344 , see figure [ fig:2horoballs ] ( b ) .
let @xmath345 , and define @xmath346 and @xmath347 . with the techniques of proposition [ proposition : s4 ] and horosphere equations and , we compute that @xmath348 and @xmath349 .
the corner of the simplex at @xmath312 is half the size of that at @xmath333 so we have that @xmath350 when @xmath351 . by lemma [ lemma : szirmai ] @xmath352 which is maximal when @xmath134 is maximized , this happens when @xmath353 . @xmath354 the data for the optimal horoball packing is summarized in table [ table : data_mult ] .
the symmetry group @xmath355 carries the density from the fundamental domain to the entire tiling .
similarly to the above proof we obtain the optimal horoball packing density for coxeter simplex tiling @xmath356 is @xmath357 .
note that @xmath358 so the tilings are related by fundamental domain doubling .
the optimal horoball configurations of @xmath166 and @xmath169 are essentially the same .
we generalize the above results to the two triply asymptotic tilings using the subgroup relations of the multiply asymptotic tilings given in figure [ fig : lattice_of_subgroups ] .
the indices of the subgroups are @xmath359 for all multiply asymptotic cases . by repeated use of lemma [ lemma : szirmai ] ,
we can show that this value is the optimal packing density for all multiply asymptotic cases .
we omit the technical details of the proof . the optimal horoball packing density for triply asymptotic coxeter simplex tilings @xmath218 , @xmath360 is @xmath220 .
a summary of the results for multiply asymptotic tilings are given in table [ table : data_mult ] .
& & + witt symb . &
@xmath166 & @xmath169 & @xmath172 & @xmath174 + + @xmath122 & @xmath361 & @xmath361 & @xmath361 & @xmath361 + @xmath222 & @xmath362 & @xmath362 & @xmath363 & @xmath363 + @xmath228 & @xmath364 & @xmath365 & @xmath364 & @xmath365 + @xmath234 & @xmath366 & @xmath367 & @xmath367 & @xmath367 + @xmath239 & @xmath225 & @xmath225 & @xmath225 & @xmath225 + + @xmath245 & @xmath368 & @xmath368 & @xmath369 & @xmath369 + @xmath251 & @xmath370 & @xmath370 & @xmath370 & @xmath370 + @xmath257 & @xmath371 & @xmath371 & @xmath371 & @xmath371 + @xmath263 & @xmath372 & @xmath372 & @xmath372 & @xmath372 + @xmath269 & @xmath270 & @xmath373 & @xmath270 & @xmath373 + + @xmath374 & @xmath275 & @xmath275 & @xmath375 & @xmath375 + @xmath376 & @xmath377 & @xmath377 & @xmath378 & @xmath378 + @xmath379 & @xmath378 & @xmath378 & @xmath378 & @xmath378 + + @xmath380 & @xmath299 & @xmath300 & @xmath299 & @xmath300 + @xmath381 & @xmath377 & @xmath377 & @xmath299 & @xmath300 + @xmath382 & @xmath383 & @xmath384 & @xmath384 & @xmath385 + + @xmath386 & 0.71644896 & 0.71644896 & 0.71644896 & 0.71644896 +
in this paper we have concentrated on horoball packings of asymptotic coxeter simplex tilings of @xmath148 . the main result of this paper is summarized in the following theorem : [ mainresult ] in @xmath148 the horoball packing density @xmath387 is optimal in seven asymptotic coxeter simplex tilings @xmath388 , when horoballs of different types are allowed at each asymptotic vertex of the tiling . the optimal packing density obtained in theorem [ mainresult ] is the densest ball packing of @xmath148 known to the authors at the time of writing . it is greater than the value @xmath389 conjectured by l. fejes tth as the realizable packing density upper bound , pp .
323 @xcite .
however , it does not exceed the brczky - type upper bound for @xmath148 of @xmath59 .
the packings we described give a new lower bound for the optimal ball packing density of @xmath148 .
in this paper we considered the generalized simplicial densities of the horoball packings .
it would be instructive to compare to the local dirichlet voronoi densities of each horoball in our family of packings , and present the density of the packing as a weighted average over the cells .
results on the monotonicity of simplicial density function @xmath43 for @xmath391 may help establish the optimality of our packings in @xmath148 as in the case of @xmath51 ( cf .
section 1 ) .
these questions are the subject of ongoing research .
bezdek , k. sphere packings revisited , _
european journal of combinatorics _ , * 27/6 * ( 2006 ) , 864883 .
bowen , l. - radin , c. densest packing of equal spheres in hyperbolic space , _ discrete and computational geometry _
, * 29 * ( 2003 ) , 2339 .
brczky , k. packing of spheres in spaces of constant curvature , _ acta math .
_ , * 32 * ( 1978 ) , 243261 .
brczky , k. - florian , a. ber die dichteste kugelpackung i m hyperbolischen raum , _ acta math .
_ , * 15 * ( 1964 ) , 237245 . hales , t. c. historical overview of the kepler conjecture , _ discrete and computational geometry _ , * 35 * ( 2006 ) , 520
. fejes tth , g. - kuperberg , w. packing and covering with convex sets , handbook of convex geometry volume b , eds .
gruber , p.m. , willis j.m .
799 - 860 , _ north - holland _ , ( 1983 ) .
fejes tth , g. - kuperberg , g. - kuperberg , w. highly saturated packings and reduced coverings , _ monatshefte fr mathematik _ , * 125/2 * ( 1998 ) , 127145 .
fejes tth , l. regular figures , _ macmillian ( new york ) _ , 1964 .
jacquemet , m. the inradius of a hyperbolic truncated @xmath3-simplex _ discrete and computational geometry _ , * 51/4 * ( 2014 ) , 9971016 .
johnson , n.w . , kellerhals , r. , ratcliffe , j.g . ,
tschants , s.t .
the size of a hyperbolic coxeter simplex , _ transformation groups _ , * 4/4 * ( 1999 ) , 329353 .
johnson , n.w .
, kellerhals , r. , ratcliffe , j.g . , tschants , s.t
commensurability classes of hyperbolic coxeter groups , _ linear algebra and its applications _ , * 345 * ( 2002 ) , 119147 .
johnson , n.w . ,
weiss , a.i .
quaternionic modular groups , _ linear algebra and its applications _ , * 295/1 - 3 * ( 1999 ) , 159189 .
kellerhals , r. the dilogarithm and volumes of hyperbolic polytopes , _ ams mathematical surveys and monographs _ , * 37 * ( 1991 ) , 301336 . kellerhals , r. ball packings in spaces of constant curvature and the simplicial density function , _ journal fr reine und angewandte mathematik _ , * 494 * ( 1998 ) , 189203 .
kozma , r.t . , szirmai , j. optimally dense packings for fully asymptotic coxeter tilings by horoballs of different types , _ monatshefte fr mathematik _ , * 168/1 * ( 2012 ) , 2747 .
marshall , t. h. asymptotic volume formulae and hyperbolic ball packing , _ annales academic scientiarum fennica : mathematica _ , * 24 * ( 1999 ) , 3143 .
molnr , e. the projective interpretation of the eight 3-dimensional homogeneous geometries , _ beitr .
algebra geom . , _ , * 38/2 * ( 1997 ) , 261288 .
symmetries in the 8 homogeneous 3-geometries , _ symmetry cult .
_ , * 21/1 - 3 * ( 2010 ) , 87 - 117 .
radin , c. the symmetry of optimally dense packings , non - eucledian geometries , eds .
a. prkopa , e. molnr , pp .
197 - 207 , _ springer verlag _ , ( 2006 ) .
rogers , c.a .
packing and covering , cambridge tracts in mathematics and mathematical physics 54 , _ cambridge university press _ , ( 1964 ) .
szirmai , j. the optimal ball and horoball packings of the coxeter tilings in the hyperbolic 3-space _ beitr .
algebra geom . ,
_ * 46/2 * ( 2005 ) , 545558 .
szirmai , j. the optimal ball and horoball packings to the coxeter honeycombs in the hyperbolic d - space _ beitr .
algebra geom . ,
_ * 48/1 * ( 2007 ) , 3547 .
szirmai , j. the densest geodesic ball packing by a type of nil lattices , _ beitr .
algebra geom .
, _ * 48/2 * ( 2007 ) , 383397 . szirmai , j. the densest translation ball packing by fundamental lattices in sol space , _ beitr .
algebra geom . ,
_ * 51/2 * ( 2010 ) , 353373 .
szirmai , j. horoball packings to the totally asymptotic regular simplex in the hyperbolic n - space , _ aequationes mathematicae _ , * 85 * ( 2013 ) , 471 - 482 , doi : 10.1007/s00010 - 012 - 0158 - 6 .
szirmai , j. horoball packings and their densities by generalized simplicial density function in the hyperbolic space , _ acta math . hung .
, _ * 136/1 - 2 * ( 2012 ) , 3955 , doi : 10.1007/s10474 - 012 - 0205 - 8 .
szirmai , j. regular prism tilings in @xmath392 space , _ aequationes mathematicae _ , ( 2013 ) , doi : 10.1007/s00010 - 013 - 0221-y .
szirmai , j. simply transitive geodesic ball packings to @xmath393 space groups generated by glide reflections , _ annali di matematica pura ed applicata _ , ( 2013 ) , doi : 10.1007/s10231 - 013 - 0324-z . | in this paper we consider ball packings in @xmath0-dimensional hyperbolic space . we show that it is possible to exceed the conjectured @xmath0-dimensional realizable packing density upper bound due to l. fejes - tth ( regular figures , 1964 ) .
we give seven examples of horoball packing configurations that yield higher densities of @xmath1 , where horoballs are centered at ideal vertices of certain coxeter simplices , and are invariant under the actions of their respective coxeter groups . |
one of the most vexing problems in the classification of surfaces of general type is the fact that we are not yet able to classify surfaces with the smallest possible invariants @xmath3 and @xmath2 .
such surfaces are usually called ( numerical ) godeaux surfaces @xcite .
nowadays the moduli space of surfaces of general type comes with a modular compactification , the moduli space of stable surfaces @xcite .
the stable surfaces occurring in the compactification can be used to gain insight in the classical moduli space but are also interesting in their on right .
this article grew out of two motivations : first of all we hoped to give a new construction of some godeaux surfaces by smoothing stable godeaux surfaces in the spirit of @xcite but starting with a gorenstein surface .
secondly , in our study of pluricanonical maps of stable surfaces @xcite we were looking for examples of gorenstein stable surfaces with @xmath4 not very ample ; so it was natural to study some examples with small invariants .
both motivations lead us to the construction described in section [ sect : surface ] where from any general smooth del pezzo surface of degree 1 we construct a ( gorenstein ) stable godeaux surface , that is a gorenstein stable surface @xmath0 with @xmath1 and @xmath2 .
we also show that @xmath0 is topologically simply connected and compute its integral homology .
it turns out that our first hope was futile : the surfaces we construct are far away from any smooth surfaces .
let @xmath0 be a stable godeaux surface of the type constructed in section [ sect : surface ] .
then the kuranishi space of @xmath0 is smooth of dimension 8 and all deformations are locally trivial .
such surfaces form an irreducible open subset of the moduli space @xmath5 of stable surface with @xmath6 , which has at most finite quotient singularities .
the proof of theorem a will be given in section [ sect : proof ] . the key point is the explicit control over infinitesimal deformations in section [ sect : deformations ] , where we have to overcome the annoying fact that we are handling local normal crossing surfaces which do not have global normal crossings .
in section [ sect : 3 ] we will concentrate on one explicit example and compute , with the help of a computer algebra system , its canonical ring . the result is a rather complicated ring , gorenstein in codimension ten , but it is generated in degree up to 5 and , in particular , @xmath4 is very ample .
in fact , one can show directly that the latter property is true for all surfaces in our family @xcite . in principle
, it would be possible to study degenerations of our examples via the del pezzo surfaces , however it is unclear if in this way one would be able to connect to a known component of the moduli space of godeaux surfaces .
starting from the results in @xcite one can actually classify gorenstein stable godeaux surfaces with worse than canonical singularities .
these will be studied in a forthcoming joint paper with marco franciosi and rita pardini .
we are indebted to marco franciosi , rita pardini , and wenfei liu for many discussions on stable surfaces .
we also enjoyed discussing different aspects of this project with olivier benoist , anne frbis - krger , roberto pignatelli , miles reid , helge ruddat , and timo schrg .
the work of the author is supported by the dfg through the emmy noether program and sfb 701 .
part of this research was carried out while visiting the university of pisa , supported by gnsaga of indam , and during a stay at him in bonn .
[ construction ] let @xmath7 be a smooth del pezzo surface of degree 1 , that is , a smooth projective surface with @xmath8 ample and @xmath9 . assume that there are two different curves @xmath10 such that @xmath11 is isomorphic to a plane nodal cubic .
we will see below that this condition is satisfied for a general choice of @xmath7 .
let @xmath12 and let @xmath13 be the normalisation map .
we specify an involution @xmath14 on @xmath15 , which preserves the preimages of the nodes , and construct a surface @xmath0 as the pushout in the diagram @xmath16{\bar\iota } & \bar d\dar{\pi } & \bar d^\nu \lar[swap]{\bar\nu}\dar{/\tau } \\ x\rar[hookleftarrow]{\iota } & d & d^\nu\lar[swap]{\nu } \end{tikzcd}.\ ] ] since @xmath15 is the disjoint union of two copies of @xmath17 , each with three marked points that are preimages of nodes of @xmath18 , the involution @xmath14 is uniquely determined by its action on these points .
this and some further notation is given in figure [ fig : construction ] .
[ curve/.style = thick , every loop/.style = looseness=10 , min distance=30 , q1/.style = color = plum , q2/.style = color = redorange , q3/.style = color = midnightblue , scale = 0.7 ] ( q3 ) at ( 0,2.3 ) ; ( q2 ) at ( 1.6,1.2 ) ; ( q1 ) at ( -1.6,1.2 ) ; ( -2.5 , -1.5 ) rectangle ( 2.5 , 3.5 ) node [ above right ] @xmath7 ; ( -2,-1 ) node [ right ] @xmath19 to[out=45 , in=315 ] ( -1.6 , 1.2 ) to[out=135 , in = 170,loop ] ( ) to[out=350 , in = 260 ] ( .2,3 ) ; ( -2,-1 ) node [ left ] @xmath20 to[out=45 , in=315 ] ( -1.6 , 1.2 ) to[out=135 , in = 170,loop ] ( ) to[out=350 , in = 260 ] ( .2,3 ) ; ( q3 ) circle ( 2pt ) ; at ( q3 ) [ right ] @xmath21 ; ( q2 ) circle ( 2pt ) ; at ( q2 ) [ above left ] @xmath22 ; ( q1 ) circle ( 2pt ) ; at ( q1 ) [ above right ] @xmath23 ; ( 0,-2 ) to node[right ] @xmath24++(0,-1 ) ; ( -2.5 , -1.5 ) rectangle ( 2.5 , 2 ) node [ above right ] @xmath0 ; ( 1,1.5 ) node [ right]@xmath25 to[out = 250 , in = 30 ] ( 0,0 ) to[out = 210 , in = 145 , loop ] ( ) to[out = 325 , in = 355 , loop ] ( ) to[out = 175 , in = 5 ] ( 179:1.5 cm ) ; ( 0,0 ) circle ( 2pt ) node [ below , yshift = -.1 cm ] @xmath26 ; ( 3,1 ) ( 2.5,1 ) node[above ] @xmath27 ( 1,1 ) node[above ] @xmath28 to ( -0.5,1 ) node[above ] @xmath29(-1,1 ) ; ( 2.5,1 ) circle ( 2pt ) ; ( 1,1 ) circle ( 2pt ) ; ( -0.5,1 ) circle ( 2pt ) ; ( 3,-1)(2.5,-1 ) node[below ] @xmath30(1,-1 ) node[below ] @xmath31(-0.5,-1 ) node[below ] @xmath32(-1,-1 ) ; ( 2.5,-1 ) circle ( 2pt ) ; ( 1,-1 ) circle ( 2pt ) ; ( -0.5,-1 ) circle ( 2pt ) ; at ( 3.5 , 2 ) @xmath15 ; ( 2.5 , 0.5 )
( 2.5 , -0.5 ) ; ( 1 , 0.5 ) ( 1 , -0.5 ) ; ( -.5 , 0.5 )
( -.5 , -0.5 ) ; at ( 3,0 ) @xmath14 ; ( -2.5,0 ) to node[above ] @xmath33++(-1,0 ) ; ( 1,-3 ) to node[right ] @xmath34++(0,-1 ) ; ( 3,0 ) (2.5,0 ) circle ( 2pt ) node[above ] @xmath35(1,0 ) circle ( 2pt ) node[above ] @xmath36(-0.5,0 ) circle ( 2pt ) node[above ] @xmath37(-1,0 ) ; at ( 3.5 , 1 ) @xmath38 ; ( -2.5,0 ) to node[above ] @xmath39++(-1,0 ) ; note that @xmath40 , which is ample . by (
* thm . 5.13 )
the surface @xmath0 exists as a projective scheme , has semi - log - canonical singularities and @xmath41 is ample .
in fact , in our case it has a degenerate cusp singularity at @xmath26 , locally isomorphic to the origin of @xmath42 , normal crossings along @xmath43 and is smooth outside @xmath25 .
[ prop : invariants ] the surface @xmath0 has semi - log - canonical , gorenstein singularities and @xmath41 is ample .
it has invariants @xmath44 and thus is a stable godeaux surface .
it is topologically simply connected and has integral homology @xmath45 the statement on singularities and ampleness was proved above and we have @xmath46 as in ( * ? ? ?
14 ) we also have @xmath47 and thus it remains to show that @xmath48 .
so assume we have a non - zero effective divisor @xmath49 .
then @xmath50 which is an elliptic pencil with base point @xmath21 .
thus @xmath51 contains @xmath21 which means that @xmath25 itself contains the degenerate cusp @xmath26 .
consequently @xmath52 contains @xmath23 and @xmath22 as well , which is impossible for a divisor in @xmath53 .
thus @xmath54 and consequently also @xmath55 .
[ sphere/.style = thin , graph/.style = thick , fill = black , radius = 2.5pt , scale = .8 ] ( -2,0 ) circle node[above ] @xmath29++ ( 1,0 ) circle node[above ] @xmath28++ ( 1,0 ) circle node[above ] @xmath27++(1,0 ) ; ( 1.4,0 ) circle [ radius = .4 ] circle [ x radius=.4 cm , y radius=.1cm]++(.4,0 ) node[right ] @xmath19 ; ( -2,-2 ) circle node[above ] @xmath32++ ( 1,0 ) circle node[above ] @xmath31++ ( 1,0 ) circle node[above ] @xmath56++(1,0 ) ; ( 1.4,-2 ) circle [ radius = .4 ] circle [ x radius=.4 cm , y radius=.1 cm ] + + ( .4,0 ) node[right ] @xmath20 ; ; ( 3,0 ) to node[right]@xmath14 ( 3,-2 ) ; ( 0,0 ) coordinate(q1 ) + + ( 2,0 ) coordinate ( q3 ) + + ( 0,1 ) coordinate(q2 ) ; ( q1 ) + + ( 0,-1.4 ) circle [ radius = .4 ] circle [ x radius=.4 cm , y radius=.1cm]++(.4,0 ) node[right ] @xmath19 ; ( q1 ) ++(0,-1 ) ; ( q3 ) + + ( 0,-1.4 ) circle [ radius = .4 ] circle [ x radius=.4 cm , y radius=.1cm]++(.4,0 ) node[right ] @xmath20 ; ( q3 ) ++(0,-1 ) ; ( q1 ) to[out=-180 , in = 90,loop]node[above left ] @xmath57 ( ) ; ( q3 ) to node[above]@xmath58 ( q1 ) ; ( q2 ) to[out=0 , in = 90,loop]node[above right ] @xmath58 ( ) ; ( q2 ) to node[right]@xmath57 ( q3 ) ; ( 8,0 ) coordinate(p ) ; ( p ) + + ( 0,-1 ) ; ( p)++(0,-1.4 ) circle [ radius = .4 ] circle [ x radius=.4 cm , y radius=.1cm]++(0.4,0 ) node[right]@xmath25 ; ( p ) to[in=15 , out = 75,loop ] node[above right]@xmath58 ( ) ; ( p ) to[out=165 , in = 105,loop ] node[above left ] @xmath57 ( ) ; ( q1 ) circle node[below left ] @xmath23 ( q2 ) circle node [ left ] @xmath22 ( q3 ) circle node[below left ] @xmath21 ( p ) circle node[below left]@xmath26 ; ( 4,0 ) to node[above]@xmath24 + + ( 1.5,0 ) ; for the computation of topological invariants we first choose a suitable homotopy - equivalent models @xmath59 : we replace each component of @xmath18 by a 2-sphere with an attached interval on which the glueing takes place ; this is shown in figure [ fig : topological model ] . in this model , the maps in the mayer - vietoris - sequence in homology ( see ( * ? ? ?
3.2 ) ) @xmath60 are easily computed and give the claimed result . for the fundamental group
we have , again by ( * ? ? ?
3.2 ) , @xmath61 because @xmath7 is simply connected .
the vanishing of the irregularity can be proved without resorting to the explicit construction given above ( see @xcite ) .
recall that a smooth del pezzo surface @xmath7 of degree 1 is the blow up of @xmath62 in eight points @xmath63 in general position @xcite . on @xmath62 the anti - canonical pencil of @xmath0
induces a pencil of cubics and , conversely , given a ( general ) pencil of cubics we can blow up eight of the base points to get a del pezzo surface of degree 1 . for later use
we record [ lem : def dp ] let @xmath64 be a smooth del pezzo surface of degree @xmath65 . then @xmath66 and @xmath67 .
the local universal deformation space of @xmath0 is smooth of dimension @xmath68 and every deformation is induced by a deformation of the points @xmath69 in the plane .
let @xmath70 be the ideal sheaf of the points in the plane .
then by ( * ? ? ?
* lem . 9.22 ) @xmath71 and there are no higher pushforward sheaves . since the points are in general position @xmath72 and the dimensions of the cohomology groups of @xmath73 follow from the exact sequence @xmath74 since @xmath75 the surface has a smooth universal deformation of dimension @xmath76 by ( * ? ? ?
the surjection @xmath77 is the map that associates to an infinitesimal deformation of the points in @xmath62 the corresponding deformation of @xmath0 , which shows the last point . blowing up all nine base points of a pencil of cubics we obtain an elliptic fibration @xmath78 with euler number @xmath79 ,
so every pencil contains singular cubics .
since cubics with worse than nodal singularities form a subset of codimension at least two in the space of cubics , the general pencil will contain 12 nodal cubics as singular elements .
in particular , the general del pezzo surface of degree 1 satisfies the assumptions made in construction [ construction ] .
we want to carry out the construction in from the previous section in a family : fix once for all coordinates @xmath80 in @xmath62 and let @xmath81 .
then let @xmath82 looking at the projection to @xmath19 we see that @xmath83 is irreducible ; the projection of @xmath83 to the locus of nodal curves in @xmath84 has finite fibres , thus @xmath85 .
note that each pair @xmath86 uniquely determines points @xmath63 such that @xmath87 .
let @xmath88 and let @xmath89 be the universal divisor given fibrewise by @xmath20 .
we blow up @xmath90 in those components of @xmath91 that do not contain the point @xmath92 in any fibre . passing to a smooth open subset @xmath93 and denoting the strict transforms of @xmath94 with the same symbol we get a flat family of pairs @xmath95 of smooth del pezzo surfaces of degree 1 together with two different nodal cubics in the anti - canonical system . possibly after a finite tale
cover , which we omit from the notation , the normalisations @xmath96 are trivial @xmath17-bundles over @xmath97 and the fibrewise involution described in construction [ construction ] extends to an involution on the family . by (
* cor . 5.33 )
, there is a family @xmath98 such that every fibre is a surface as constructed in the previous section .
note that the family of singular loci of the fibres @xmath99 is the trivial family , thus flat .
therefore in the exact sequence @xmath100 the first and the third sheaf are flat over @xmath97 , so also @xmath101 is flat .
we wrap up the discussion so far in the following statement : [ prop : family ] there are flat families @xmath102{\bar\phi } \rar & \gothx\dar{\phi}\\&b\end{tikzcd}\ ] ] over an irreducible smooth base of dimension @xmath68 such that 1 .
every fibre of @xmath101 is a gorenstein stable godeaux surface as described in section [ sect : surface ] and every general such surface occurs in the family , 2 .
@xmath103 is the simultaneous normalisation of the fibres of @xmath101 , thus a family of del pezzo surfaces of degree @xmath65 , 3 .
the family @xmath104 is universal in every point , thus @xmath97 admits a finite map to the moduli space of del pezzo surfaces of degree 1 .
we only need to show the last point . by lemma
[ lem : def dp ] a deformation of a smooth fibre @xmath7 is uniquely determined by a deformation of the 8 points in the plane that we blow up . in our construction
we assumed that the points are contained in @xmath19 , so any sufficiently small deformation of the points will still by contained in a nodal cubic . under the action of @xmath105 all nodal cubics are equivalent ,
so all deformations of @xmath7 are in fact induced by moving the points on @xmath19 , which is what is parametrised by @xmath97 .
the universality now follows by smoothness and dimension reasons .
to understand the deformations of the surfaces constructed above we start in a slightly more general setting : let @xmath0 be a surface with at most local normal crossing singularities .
we denote by @xmath25 the double locus in @xmath0 , by @xmath7 the normalisation and by @xmath18 the preimage of @xmath25 in @xmath7 .
then @xmath18 is a nodal curve and the glueing induces an involution @xmath14 on @xmath15 , the normalisation of @xmath18 .
the involution preserves preimages of nodes and we have a pushout diagram as in .
we denote as usual is a local complete intersection , even locally a hypersurface , so we can replace the cotangent complex by the sheaf of khler differentials . ]
@xmath106 such that @xmath107 is the tangent sheaf of @xmath0 and @xmath108 is the space of infinitesimal deformations of @xmath0 .
the local - to - global-@xmath109-sequence ( see ( * ? ? ?
2.4.1 ) ) reads as @xmath110 because @xmath0 is a local complete intersection and thus @xmath111 . to analyse the spaces in the above sequence we use some of friedman s results from @xcite , which we will now recall .
the natural map @xmath112 has as image @xmath113 , the double dual of @xmath114 , and as kernel the sheaf of torsion differentials @xmath115 . by (
1.5 ) we have two short exact sequences @xmath116 furthermore ( * ? ? ?
1.10 ) there are line bundles @xmath117 and @xmath118 such that @xmath119 is exact .
we need to compute the degree of these line bundles , so we look at local generators : near a normal crossing point of @xmath0 with local equation @xmath120 the sequence corresponds to @xmath121 while at the degenerate cusp with equation @xmath122 we have @xmath123 on one ( local ) component of the normalisation , say @xmath124 , the function @xmath125 , resp .
the form @xmath126 is a generator for the conormal bundle of @xmath127 in @xmath128 ; writing @xmath129 we see that a generator of the conormal bundle of @xmath130 is a generator of @xmath131 . by ( * ? ? ? *
prop.2.3 ) the dual of @xmath117 is @xmath132 and thus by patching together the above local computation we get : [ lem : degree t1 ] let @xmath14 be the involution acting on @xmath15 , the normalisation of @xmath18 and let @xmath133 be the preimages of nodes of @xmath18 .
then @xmath134 is locally free and @xmath135 is the line bundle on @xmath136 corresponding the following @xmath14-invariant line - bundle on @xmath15 : @xmath137 we now turn back to our original example .
[ prop : step 1 ] in the situation of construction [ construction ] the following hold true : 1 .
the inclusion @xmath138 induces for all @xmath139 isomorphisms in cohomology @xmath140 .
2 . @xmath141 and the local - to - global-@xmath109-sequence becomes @xmath142 3 .
@xmath143 and if both groups vanish then @xmath0 has unobstructed deformations and all deformations of @xmath0 are locally trivial deformations induced by a deformation of the del pezzo surface @xmath7 .
for we have to show that @xmath144 , the cokernel of the inclusion , has no cohomology . dualising the restriction sequence of khler differentials we get two exact sequences ( * ? ? ?
1.1.8 ) @xmath145 we interpret the cohomology groups of these sequences in deformation theoretic terms ( * ? ? ?
3 ) : * @xmath146 is the tangent space to the automorphism group of @xmath18 , thus vanishes since @xmath18 is stable . *
@xmath147 is the space of infinitesimal locally trivial deformations of @xmath18 , thus vanishes .
* @xmath148 is the map that associates to an infinitesimal deformations of @xmath18 inside @xmath7 the local deformation at the nodes of @xmath18 .
both spaces have dimension @xmath149 , the first by the restriction sequence and the second because @xmath18 has three nodes .
the map @xmath150 is an isomorphism , because the nodes of @xmath18 can be smoothed independently in @xmath7 : deforming @xmath11 to a smooth canonical curve smoothed @xmath151 while keeping the other two nodes and deforming @xmath18 to a general member of @xmath152 smoothed all nodes . *
@xmath153 and @xmath154 have no higher cohomology , thus by the second sequence and the previous item @xmath155 for all @xmath139 .
* @xmath146 is the tangent space to the automorphism group of @xmath18 , thus vanishes since @xmath18 is stable .
* @xmath147 is the space of infinitesimal locally trivial deformations of @xmath18 , thus vanishes .
thus by the first sequence @xmath156 for all @xmath139 as claimed .
we now turn to . since @xmath157 the sequence induces an isomorphism @xmath158 .
note that the relative dualising sheaf for @xmath159 is @xmath160 and similarly the relative dualising sheaf for @xmath161 is @xmath162 .
hence by relative duality @xmath163 and thus applying @xmath164 to yields a short exact sequence @xmath165 in particular , we have induced isomorphisms @xmath166 and the our claim follows from the first item . for note
that by lemma [ lem : degree t1 ] the line bundle @xmath134 on @xmath25 has degree @xmath167 .
since a line bundle on an irreducible curve of arithmetic genus @xmath168 can have at most one section , we get the first part of . if @xmath169 then , by , we have natural isomorphism @xmath170 and @xmath171 , and our claim follows from (
2.6.4 ) . next
we show that only the favourable case in the last item of the previous proposition occurs .
[ lem : step 2 ] in the situation of construction [ construction ] we have @xmath172 .
assume there is a non - zero section @xmath173 . since @xmath132 has degree 1 on @xmath25 , the section @xmath174 vanishes exactly in one smooth point @xmath175 . by proposition [ prop :
step 1 ] this section lifts to an infinitesimal deformation @xmath176/(\epsilon^2)\right)$ ] of @xmath0 .
locally analytically along the singular locus of @xmath0 , the family looks like the following models in @xmath177 : @xmath178 we will now derive a contradiction as follows : guided by the local models we construct a partial - log - resolution of the central fibre @xmath179 with central fibre @xmath180 . using the description of @xmath181 from lemma [ lem : degree t1 ]
we then show that @xmath180 can not have a deformation which is non - singular in codimension 1 .
[ curve/.style = thick , every loop/.style = looseness=10 , min distance=30 , q1/.style = color = plum , q2/.style = color = redorange , q3/.style = color = midnightblue , scale = 0.7 ] ( q3 ) at ( 0,2.3 ) ; ( q2 ) at ( 1.6,1.2 ) ; ( q1 ) at ( -1.6,1.2 ) ; ( -2.5 , -1.5 ) rectangle ( 2.5 , 3.5 ) node [ above right ] @xmath7 ; ( -2,-1 ) node [ right ] @xmath19 to[out=45 , in=315 ] ( -1.6 , 1.2 ) to[out=135 , in = 170,loop ] ( ) to[out=350 , in = 260 ] ( .2,3 ) ; ( -2,-1 ) node [ left ] @xmath20 to[out=45 , in=315 ] ( -1.6 , 1.2 ) to[out=135 , in = 170,loop ] ( ) to[out=350 , in = 260 ] ( .2,3 ) ; ( q3 ) circle ( 2pt ) ; at ( q3 ) [ right ] @xmath21 ; ( q2 ) circle ( 2pt ) ; at ( q2 ) [ above left ] @xmath22 ; ( q1 ) circle ( 2pt ) ; at ( q1 ) [ above right ] @xmath23 ; ( 0,-2 ) to node[right ] @xmath24++(0,-1 ) ; ( -2.5 , -1.5 ) rectangle ( 2.5 , 2 ) node [ above right ] @xmath0 ; ( 1,1.5 ) node [ right]@xmath25 to[out = 250 , in = 30 ] ( 0,0 ) to[out = 210 , in = 145 , loop ] ( ) to[out = 325 , in = 355 , loop ] ( ) to[out = 175 , in = 5 ] ( 179:1.5 cm ) ; ( 0,0 ) circle ( 2pt ) node [ below , yshift = -.1 cm ] @xmath26 ; ( -2.5 , -1.5 ) rectangle ( 2.5 , 3.5 ) node [ above right ] @xmath182 ; ( -1.6,-2 ) rectangle ( 0 , 2 ) ; at ( -3,.5 ) [ circle through=(-1.5,-.5 ) ] ; at ( -.8 , .5 ) @xmath183 ; ( 1.6,-2 ) rectangle ( 0 , 2 ) ; at ( 3,.5 ) [ circle through=(1.5,-.5 ) ] ; at ( .8 , .5 ) @xmath184 ; ( -1.6,2.5 ) to node [ below ] @xmath185 + + ( 3.2,0 ) ; ( -1.5,-1 ) node [ left ] @xmath186 + + ( 0,4 ) ; ( 1.5,-1 ) node [ right ] @xmath187 + + ( 0,4 ) ; \(a ) at ( -90:2 ) ; ( b ) at ( 30:2 ) ; ( c ) at ( 150:2 ) ; ( a)(b)(c) cycle ; ( a )
node[below left ] @xmath127(c ) ; ( a )
node[below right ] @xmath188(b ) ; ( b ) node[above ] @xmath189 ( c ) ; at ( 0,0 ) @xmath62 ; at ( 2.5,1)@xmath190 ; ( 3,-2 ) to node[left ] @xmath191node [ right]@xmath192 + + ( 0,-1 ) ; ( -2.5,1 ) to node[above ] @xmath193 + + ( -1,0 ) ; ( -2.5 , -1.75 ) rectangle ( 2.5 , 1.75 ) node [ above right ] @xmath180 ; \(a ) at ( -90:1 ) ; ( b ) at ( 30:1 ) ; ( c ) at ( 150:1 ) ; ( a)(b)(c) cycle ; ( a ) ( c ) ; ( a ) ( b ) ; ( b ) (c ) ; at ( 0,0 ) @xmath62 ; ( -120:1 ) arc ( -120:180:1 ) ; at ( -30:1.5 ) @xmath136 ; ( -4,0 ) to node[above ] @xmath193 + + ( -2,0 ) ; so we blow up @xmath194 . from the local model
we see that the normalisation of the central fibre , which is again a normal crossing divisor , consists of two components : @xmath195 these are glued together as in figure [ fig : log - resolution ] and the singular locus has four irreducible components : the curve @xmath136 , which is the strict transform of the singular locus in @xmath0 and three curves @xmath196 where we glue an exceptional curve of the blow up @xmath197 to a line in @xmath62 . by lemma [ lem : degree t1 ]
we compute @xmath198 thus every section of @xmath199 vanishes along @xmath200 which is a contradiction to the existence of the family @xmath201 constructed .
i would have liked to compute @xmath202 directly in terms of the construction , instead of constructing the partial log - resolution .
however , the precise way to nail down the glueing data on @xmath203 which makes it a sheaf on @xmath25 evaded me .
it seems plausible that some of the above can be formulated favourably in the language of log - structures but it is unclear to me if this gives a neater proof .
in section [ sect : family ] we constructed a family @xmath204 of gorenstein stable surfaces over a smooth base which parametrises the general surface as constructed in section [ sect : surface ] .
in addition , the family admits a simultaneous normalisation which is a family of del pezzo surfaces of degree 1 , universal in every point ( see proposition [ prop : family ] ) .
we have proved theorem @xmath205 as soon as we show that this family is locally at a point @xmath206 the universal deformation space of the fibre @xmath207 .
this is accomplished by proposition [ prop : step 1 ] combined with lemma [ lem : step 2 ] .
one original motivation to study this example was the search for gorenstein stable surfaces with a complicated canonical ring and possibly the 5-canonical map not an embedding .
we thought that this example was suitable , because the invariants are small and the double curve prevents the 4-canonical map from being an embedding . more precisely , we proved in ( * ? ? ? * example 47 ) that since @xmath41 has degree 1 one @xmath25 the 2-canonical map has base points and neither the 3-canonical nor the 4-canonical map are embeddings when restricted to @xmath25
. we will disprove these expectations in one particular example , that we will now describe .
our computation , which we will only sketch , is based on the following result of kollr , which we state in a simplified version .
then a section @xmath210 descends to a section in @xmath211 if and only if the image of @xmath212 in @xmath213 under the residue map is @xmath14-invariant if @xmath214 is even respectively @xmath14-anti - invariant if @xmath214 is odd .
let @xmath215 $ ] be a polynomial ring with weights @xmath216 and let @xmath217 .
then @xmath218 is a smooth del pezzo surface of degree 1 , @xmath219 is the anti - canonical ring of @xmath7 and the equations @xmath220 define nodal rational curves @xmath221 .
we now construct a gorenstein stable godeaux surface as in construction [ construction ] . in the notation of figure
[ fig : construction ] we have @xmath223 which has degree one on each component .
we choose generators for @xmath224 , the ring of sections of @xmath225 , such that the sum of the residue map in the single degrees is described algebraically by @xmath226\times \ic[a_2,b_2 ] \\
( x_1 , x_2 , y , z)\mapsto & \ , ( b_2-a_2 , b_1-a_1 , -a_1b_1-a_2b_2 , -a_1 ^ 2b_1+a_2 ^ 2b_2 ) \end{split}\ ] ]
let @xmath227 be the subring of @xmath224 consisting of @xmath14-anti - invariant sections in odd degrees and @xmath14-invariant sections in even degrees , that is , the pullback to @xmath15 of the ring of sections of @xmath228 .
by proposition [ prop : sections ] the canonical ring of @xmath0 is the preimage of @xmath227 in @xmath219 . for a particular choice of @xmath14
, we have @xmath229\subset r_{\bar d^\nu}.\ ] ] 1 .
the canonical ring @xmath231 is generated as a subring of @xmath219 by @xmath232 and there are 54 relation of in degrees @xmath233 . consequently , there is an embedding into weighted projective space @xmath234 as a subvariety of codimension ten . 2 . the bicanonical map @xmath235 has a unique base point at @xmath26
3 . the tricanonical map is a birational morphism onto the surface in @xmath236 of degree 9 with equation @xmath237 the singular locus consists of the two lines @xmath238 and @xmath239 ; the first line is the image of @xmath25 .
a real picture of the surface with the singular locus clearly visible is shown in figure [ fig : x3 ] .
+ 1 . a very successful method to study canonical rings is restriction to a canonical curve . here
we do nt have a canonical curve but it is tempting to look at the restriction to the non - normal locus @xmath25 .
however , this does not help , because @xmath25 is not cartier and in fact there is a generator in degree 5 that vanishes on @xmath25 .
the properties of the pluricanonical maps in low degrees can easily be worked out directly without the help of a computer algebra system .
appearantly , the number of relations in the ring suggests that it might be obtained by repeated unprojections but at this level of complexity , gorenstein in codimension 10 , i have not even tried to guess a general format for the canonical rings of the surfaces in the family .
jans kollr .
moduli of varieties of general type . in g.
farkas and i. morrison , editors , _ handbook of moduli : volume ii _
, volume 24 of _ advanced lectures in mathematics _ , pages 131158 . international press , 2012 , arxiv:1008.0621 . | we construct from a general del pezzo surface of degree 1 a gorenstein stable surfaces @xmath0 with @xmath1 and @xmath2 .
these surfaces are not smoothable but give an open subset of an irreducible component of the moduli space of stable godeaux surfaces . in a particular example
we also compute the canonical ring explicitly and discuss the behaviour of pluricanonical maps . |
in this paper we are interested in maximal parabolic regularity for non - autonomous parabolic equations like @xmath2 for almost every @xmath3 , where @xmath4 , @xmath5 and the operators @xmath6 all have the same domain of definition @xmath7 in a banach space @xmath8 .
if the operator function @xmath9 is constant , these equations may be solved using the concept of maximal parabolic regularity , see for example @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite .
this theory extends to cases in which the dependence @xmath10 is continuous , see ( * ? ? ?
* chapter vi ) , @xcite , @xcite and the survey in @xcite and it is a powerful tool for solving corresponding nonlinear equations , see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite .
if the continuity of is violated , things are much less understood , the only classical exception being the case that the summabilility exponent in time , @xmath11 , equals @xmath12 and that @xmath8 and @xmath7 are hilbert spaces ( see proposition [ p - lions ] below ) . for results in the banach space case
, we refer to @xcite and @xcite and for relevant recent achievements see @xcite , ( * ? ? ?
* proposition 1.3 ) , @xcite , @xcite , @xcite , @xcite and the references therein .
these results are mostly to be seen as perturbation results , with respect to an _ autonomous _ parabolic operator .
the spirit of our paper is perturbative in a different sense : not the operator is changed , but the banach space .
this enables us to extrapolate maximal parabolic regularity for whole classes of non - autonomous operators simultaneously . remarkably , grger proved in @xcite that maximal parabolic @xmath13-regularity for second - order divergence - form operators is preserved in case of non - smooth , time - dependent coefficients , if one deviates from @xmath14 to temporal and spatial integrability exponents @xmath15 in an interval @xmath16 that depends on the ellipticity constant and the @xmath17-norm of the coefficient function . here
, we provide an abstract extrapolation strategy that includes general non - autonomous operators corresponding to sesquilinear forms , and we extend the results of grger to indices @xmath18 and more general geometric settings for mixed boundary conditions .
in the first part of the paper , we develop an abstract framework which allows to show maximal parabolic @xmath19-regularity for _ non - autonomous _ operators in a banach space @xmath8 and for some @xmath20 , provided that one knows maximal parabolic regularity for suitable _ autonomous _ operators . later on
we specify @xmath8 to be @xmath21 ( see definition [ d-1 ] below ) and @xmath6 to be a second - order divergence operator @xmath22 .
we aim at situations in which the map in is substantially discontinuous , which means that it is allowed to be discontinuous in every point @xmath23 of the time interval @xmath24 , and does _ not _ satisfy the already weak condition of relative continuity in @xcite .
a prototype for this is the following : for each time @xmath25 there is a moving subdomain @xmath26 on which the coefficient function @xmath27 is constant and it takes a different constant value on @xmath28 , see section [ snonaut8 ] .
the banach spaces @xmath8 of type @xmath29 turn out to be well suited for the treatment of elliptic and parabolic problems if these are combined with inhomogeneous neumann boundary conditions ( cf .
* section 3.3 ) for @xmath30 ) or if right hand sides of distributional type appear , e.g. surface densities , which may even vary their position in time , cf .
@xcite , @xcite .
note that there is often an intrinsic connection between ( spatial ) jumps in the coefficient function and the appearance of surface densities as parts of the right hand side ( see ( * ? ? ?
* chapter 1 ) ) .
interestingly , these spaces are also adequate for the treatment of control problems , see @xcite , @xcite , @xcite , @xcite .
the advantage of @xmath29 over @xmath31 is that the domain of elliptic divergence operators continuously embeds into a hlder space if @xmath32 is larger than the space dimension ( cf .
@xcite @xcite ) , which is helpful when considering quasilinear problems , see section [ snonaut00 ] below and cf .
@xcite , @xcite , @xcite in the continuous setting .
moreover , in contrast to the @xmath33 spaces , the space @xmath29 satisfies the property that the domains of divergence operators @xmath34 coincide at different points in time even if the discontinuities of the coefficient functions @xmath27 move in @xmath35 , see the examples in section [ snonaut6 ] , taken from @xcite , @xcite .
we next give an outline of the paper .
we first recall preliminary results on maximal parabolic regularity and a quantitative version of sneiberg s extrapolation principle .
then we prove an interpolation result for different spaces of maximal parabolic regularity , i.e. we prove the interpolation identity @xmath36_\theta = l^r(j;[d , e]_\theta ) \cap w_0^{1,r}(j;[x , y]_\theta),\ ] ] in which @xmath37 and @xmath38 and @xmath39 form a _ pair of common maximal parabolic regularity _ , see definition [ d - commonpair ] below . having this at hand , one can extrapolate maximal parabolic regularity from one setting to ` neighbouring ones ' , see theorem [ t - extrapolatabstr ] .
then in section [ snonaut3 ] we treat linear , non - autonomous parabolic equations in the classical hilbert space setting as in ( * ? ? ?
* section xviii.3 ) .
we show that the exponent of time integrability extrapolates from @xmath12 to @xmath0 without losing maximal parabolic regularity and provide quantitative estimates on the size of @xmath11 in theorem [ thm - e - hilbert ] .
a detailed motivation for these type of results is given at the end of section [ snonaut3 ] . in section [ snonaut5 ]
we introduce the precise setting of the elliptic differential operators in divergence form with mixed boundary conditions which we use in the remainder of the paper . in section [ snonaut6 ]
we use recent results on elliptic regularity @xcite and autonomous parabolic regularity @xcite , @xcite to obtain maximal parabolic regularity for autonomous operators in the @xmath40-scale , even for some @xmath41 .
next in section [ snonaut7 ] we exploit in order to achieve maximal parabolic regularity for non - autonomous operators in the @xmath40-scale . to be precise , in theorem [ t - main ]
we show that non - autonomous operators @xmath42 satisfy maximal parabolic @xmath43-regularity if @xmath44 and @xmath45 .
the coefficient function @xmath46 can be discontinuous in time and space .
as in the preceding articles @xcite and @xcite , the geometric setting for the domains and boundary parts is extremely wide : the domains may even fail to be lipschitz and the ` dirichlet ' boundary part @xmath47 is only required to be ahlfors
david regular . in section [ snonaut00 ] ,
the main results from previous sections are applied to related quasilinear problems . in section [ snonaut8 ] , we give an example of a non - autonomous parabolic operator with discontinuous coefficients in both space and time to which theorem [ t - main ] applies .
throughout this paper let @xmath48 and set @xmath49 .
let us start by introducing the following ( standard ) definition .
[ d - bochner ] if @xmath8 is a banach space and @xmath50 , then we denote by @xmath19 the space of @xmath8-valued functions @xmath51 on @xmath24 which are bochner - measurable and for which @xmath52 is finite .
we define @xmath53 , where @xmath54 is to be understood as the time derivative of @xmath55 in the sense of @xmath8-valued distributions ( cf .
* section iii.1 ) ) .
moreover , we introduce the subspace @xmath56 we equip this subspace always with the norm @xmath57 . in this paper
we consider the following notion of maximal parabolic regularity in the non - autonomous case .
[ d - maxpar ] let @xmath8 , @xmath7 be banach spaces with @xmath7 densely and continuously embedded in @xmath8 .
let @xmath58 be a bounded and measurable map and suppose that the operator @xmath6 is closed in @xmath8 for all @xmath25 .
let @xmath50 .
then we say that the family @xmath59 satisfies * ( non - autonomous ) maximal parabolic @xmath19-regularity * , if for any @xmath60 there is a unique function @xmath61 which satisfies @xmath62 for almost all @xmath25 .
we write @xmath63 for the space of maximal parabolic regularity .
the norm of @xmath64 is @xmath65 then @xmath66 is a banach space .
if all operators @xmath6 are equal to one ( fixed ) operator @xmath67 , and there exists an @xmath50 such that @xmath59 satisfies maximal parabolic @xmath19-regularity , then @xmath59 satisfies maximal parabolic @xmath68-regularity for all @xmath69 and we say that @xmath67 satisfies * maximal parabolic regularity on @xmath8*. in all what follows , we denote the mapping which assigns to the right hand side @xmath60 in the solution @xmath64 by @xmath70 and @xmath71 , respectively .
the @xmath72-spaces satisfy optimal interpolation properties with respect to the complex interpolation method .
[ p - interlpp ] let @xmath73 be two banach spaces which form an interpolation couple .
further , let @xmath74 , @xmath75 and set @xmath76
. then @xmath77_\theta = l^r(j;[x , y]_\theta)\ ] ] with * equality * of norms . see ( * ? ? ? * theorem 5.1.2 ) .
we continue by quoting sneiberg s extrapolation principle .
[ t - sneib ]
let @xmath78 be banach spaces such that @xmath79 and @xmath80 are interpolation couples .
assume @xmath81 and put @xmath82 suppose that for one @xmath75 the operator @xmath83_\theta \to [ z_1,z_2]_\theta $ ] is a topological isomorphism and let @xmath84_\theta \to [ f_1,f_2]_\theta}$ ] .
then one has the following .
[ p - sneib-1 ] if @xmath85 and @xmath86 then @xmath83_{\tilde\theta } \to
[ z_1,z_2]_{\tilde\theta } $ ] remains surjective .
[ p - sneib-2 ] if @xmath85 and @xmath87 then @xmath83_{\tilde\theta } \to [ z_1,z_2]_{\tilde\theta}$ ] remains an isomorphism and @xmath88_{\tilde\theta } \to [ f_1,f_2]_{\tilde\theta } } \le 8 \beta .\ ] ] essentially , the theorem was discovered by sneiberg @xcite and elaborated in more detail in @xcite .
the explicit quantitative estimates as quoted here were worked out only recently in ( * ? ? ? * propositions 1.3.27 and 1.3.25 ) .
the aim of this section is to provide an abstract setting in which the property of non - autonomous maximal parabolic regularity can be extrapolated by using sneiberg s theorem .
[ d - commonpair ] let @xmath89 be banach spaces , with continuous embeddings @xmath90 , @xmath91 and @xmath92 .
suppose that @xmath7 is dense in @xmath8 and @xmath93 is dense in @xmath94 .
then we say that the tuples @xmath38 and @xmath39 form a * pair of common maximal parabolic regularity * if there is an operator @xmath95 such that @xmath96 , the operator @xmath97 satisfies maximal parabolic regularity on @xmath94 , and the restriction @xmath98 satisfies @xmath99 and maximal parabolic regularity on @xmath8 . for convenience ,
as is common for interpolation results , we sometimes extend the notation @xmath97 to @xmath98 or other restrictions of @xmath97 in this section . if @xmath38 and @xmath39 form a pair of common maximal parabolic regularity , then the spaces of maximal parabolic regularity interpolate as follows .
[ l - inter ] assume that @xmath38 and @xmath39 form a pair of common maximal parabolic regularity .
. then @xmath101_\theta = { { \rm mr}}^r_0(j ; [ d , e]_\theta , [ x , y]_\theta)\ ] ] for all @xmath75 , where @xmath102 .
the pair @xmath103 is an interpolation couple since both spaces continuously inject into @xmath104 .
it follows from ( * ? ? ?
* lemma 5.3 ) that @xmath97 satisfies maximal parabolic regularity on @xmath105_\theta$ ] , so @xmath106_\theta,[x , y]_\theta ) \to l^r(j;[x , y]_\theta ) \end{aligned}\ ] ] is an isomorphism .
here we used that @xmath38 and @xmath39 form a pair of common maximal parabolic regularity . on the other hand
, @xmath107 is an isomorphism from @xmath108 onto @xmath109 and from @xmath110 onto @xmath111 .
hence by interpolation , the operator @xmath112_\theta \to [ l^{r_0}(j;x ) , l^{r_1}(j;y)]_\theta , \end{aligned}\ ] ] is an isomorphism .
the statement follows from combining , and proposition [ p - interlpp ] .
this interpolation result together with theorem [ t - sneib ] enables us to extrapolate maximal parabolic regularity to non - autonomous parabolic operators .
we need a simple lemma .
[ l - measurable ] let @xmath89 be banach spaces , with continuous embeddings @xmath90 , @xmath91 and @xmath92 .
suppose @xmath7 is dense in @xmath8 and @xmath93 is dense in @xmath94 .
let @xmath113 be a subset of @xmath114 satisfying @xmath115 .
assume @xmath116 for all @xmath25 .
moreover , suppose that @xmath117 .
let @xmath75 .
then one has the following .
[ l - measurable-1.5 ] if @xmath25 , then @xmath118_\theta}\in{{\cal l}}([d , e]_\theta ; [ x , y]_\theta)$ ] . moreover @xmath119_\theta \to [ x;y]_\theta } <
\infty.\ ] ] [ l - measurable-2 ] suppose @xmath120 is measurable .
then the map @xmath121_\theta$ ] is measurable for all @xmath122_\theta$ ] .
[ l - measurable-3 ] let @xmath50 .
then the map @xmath123_\theta,[x , y]_\theta ) \to l^r(j;[x , y]_\theta)\ ] ] is well defined . `
[ l - measurable-1.5 ] ' .
this is well known from complex interpolation . `
[ l - measurable-2 ] ' . by assumption , for every @xmath124
the map @xmath125 is measurable . since the inclusion @xmath126_\theta$ ] is continuous , also the map @xmath127_\theta$ ] is measurable for all @xmath124
let now @xmath122_\theta$ ] . because @xmath7 is dense in @xmath128_\theta$ ] by (
* theorem 1.9.3 ) , there is a sequence @xmath129 in @xmath7 which converges to @xmath130 in @xmath128_\theta$ ] . but then implies that the function @xmath121_\theta$ ] is the pointwise limit of the functions @xmath131_\theta$ ] .
hence it is measurable . `
[ l - measurable-3 ] '
. one can easily deduce from statement [ l - measurable-2 ] that for every @xmath132 and @xmath133_\eta)$ ] the map @xmath134_\eta\ ] ] is measurable .
since @xmath135_\theta \to [ x;y]_\theta } < \infty$ ] , it follows that is well defined .
the main theorem of this section is the following .
[ t - extrapolatabstr ] suppose the tuples @xmath38 and @xmath39 form a pair of common maximal parabolic regularity .
for all @xmath25 let @xmath136 and suppose that @xmath116 .
assume that the maps @xmath137 are measurable and @xmath138 .
let @xmath139 and @xmath75 . set @xmath140 .
suppose that @xmath141_\theta,[x , y]_\theta ) \to l^r(j;[x , y]_\theta)\ ] ] is a topological isomorphism .
then there exists an @xmath142 such that @xmath143_{\tilde \theta},[x , y]_{\tilde \theta } ) \to l^s(j;[x , y]_{\tilde \theta})\ ] ] is a topological isomorphism for all @xmath144 , where @xmath145 the operators @xmath146 and @xmath147 are continuous and their norms are bounded by @xmath148 , where @xmath149 .
moreover , since the tuples @xmath38 and @xmath39 form a pair of common maximal parabolic regularity by assumption , we may apply lemma [ l - inter ] to obtain the interpolation identity . using also proposition [ p - interlpp ] , one can rewrite as a topological isomorphism @xmath150_\theta \to [ l^{r_0}(j;x),l^{r_1}(j;y)]_\theta .
\label{et - extrapolatabstr;1}\ ] ] by theorem [ t - sneib ] the isomorphism in remains a topological isomorphism , if @xmath151 is replaced by @xmath152 , @xmath11 is replaced by @xmath153 and @xmath154 is sufficiently close to @xmath151 .
then the theorem follows by applying again lemma [ l - inter ] and proposition [ p - interlpp ] . as a simple consequence to theorem [ t - extrapolatabstr ] , we obtain the following . [ c - extrap ]
let @xmath8 , @xmath7 be banach spaces with @xmath7 densely and continuously embedded in @xmath8 .
let @xmath155 be a bounded and measurable map and suppose that the operator @xmath156 is closed in @xmath8 for all @xmath25 .
let @xmath157 and suppose that @xmath158 satisfies maximal parabolic regularity on @xmath8 .
let @xmath50 .
suppose that @xmath159 satisfies maximal parabolic @xmath19-regularity .
then there exists an open interval @xmath160 with @xmath161 such that @xmath159 satisfies maximal parabolic @xmath68-regularity for all @xmath162
. choose @xmath163 and @xmath164 in lemma [ l - inter ] and theorem [ t - extrapolatabstr ] .
theorem [ t - extrapolatabstr ] and corollary [ c - extrap ] show the abstract principle we use in the sequel for the extrapolation of maximal @xmath19-regularity . in corollary [ c - extrap ] the space @xmath7 and @xmath8
are connected via _ some _ autonomous reference operator @xmath158 that has maximal regularity
. then corollary [ c - extrap ] gives that for every non - autonomous operator family @xmath155 with maximal parabolic @xmath19-regularity the regularity extrapolates around @xmath11 .
we expect that the interpolation formula is of independent interest and may serve for other purpose also in different contexts . in section [ snonaut3 ]
we apply this principle to non - autonomous families of operators in hilbert spaces generated by families of sesquilinear forms , and in sections [ snonaut5][snonaut7 ] to non - autonomous elliptic differential operators in sobolev spaces . in theorem [ t - extrapolatabstr ] and corollary [ c - extrap ] , however , the quantitative estimates of theorem [ t - sneib ] are lost , as holds only with equivalence of norms and the exact constants seem to be hard to control ( cf . remark
[ r - pich ] ) .
in order to get some quantitative results in the specific settings of sections [ snonaut3 ] and [ snonaut7 ] , we will use direct proofs based on theorem [ t - sneib ] .
the following definition and interpolation result will be useful tools .
[ d - tilde ] let @xmath8 , @xmath7 be banach spaces with @xmath7 densely and continuously embedded in @xmath8 .
let @xmath157 and suppose that @xmath158 satisfies maximal parabolic regularity on @xmath8 . for all @xmath50 , denote by @xmath165 the space @xmath166 equipped with the norm @xmath167 it will be clear from the context which operator @xmath158 is involved . obviously @xmath168 is an _
isomorphism . for all @xmath50
define @xmath169 then @xmath170 and @xmath171 for all @xmath64 .
so the two norms are equivalent .
[ lemma - t - intgeropo ] adopt the notation as in definition [ d - tilde ] .
let @xmath75 and @xmath139 . then @xmath172_\theta = { { \rm mr}}^r_0(j;d , x)\,\widetilde{\;}\ ] ] with * equality * of norms , where @xmath50 is such that @xmath173 . by proposition [ p - interlpp ]
one has the interpolation identity @xmath174_\theta = l^r(j;x ) , \ ] ] with equality of norms .
then the _ isometric _ isomorphism carries over to , with equality of norms .
in this section , we will investigate maximal parabolic regularity in a hilbert space setting . for convenience , recall the following classical existence result , which will serve as the starting point for the remainder of this paper .
[ p - lions ] let @xmath175 and @xmath176 be hilbert spaces with @xmath175 densely and continuously embedded into @xmath176 . for
every @xmath25 let @xmath177 be a sesquilinear form on @xmath175 .
let @xmath178 .
suppose that the map @xmath179 $ ] from @xmath24 into @xmath180 is measurable for all @xmath181 .
suppose that @xmath182 \ge c_\bullet\|\psi\|^2_v$ ] and @xmath183| \le c^\bullet \
, \|\psi\|_v \ , \|\varphi\|_v$ ] for all @xmath184 and @xmath25 .
for all @xmath25 let @xmath185 be the linear operator which is induced by the sesquilinear form @xmath177 .
then for all @xmath186 there exists a unique @xmath187 such that @xmath188 for a.e .
moreover , @xmath189 in particular , @xmath190 the existence and uniqueness in ( [ e - nonautpara ] ) follows from ( * ? ? ?
* section xviii.3 remark 9 ) .
we next prove the estimates ( [ e - apriori1 ] ) .
if @xmath191 , then the energy equality @xmath192 \ , dt = { \mathop{\rm re}}\int_0^\tau \langle f(t),u(t ) \rangle_{v^ * \times v } \ , dt\ ] ] follows from , cf .
* section xviii.3 equation 3.86 ) . using the uniform coercivity ,
this gives @xmath193 \ , dt \\
& \le & { \mathop{\rm re}}\int_0^t \langle f(t),u(t ) \rangle _ { v^ * \times v } \ , dt \\ & \leq & \int_0^t \|f(t)\|_{v^ * } \ , \|u(t)\|_v \ ,
dt \le \|f\|_{l^2(j;v^ * ) } \ , \|u\|_{l^2(j;v ) } , \end{aligned}\ ] ] which proves the first inequality in .
note that @xmath194 for all @xmath25 .
therefore @xmath195 and the second inequality in follows .
adopt the notation and assumptions of proposition [ p - lions ] .
we are interested in the problem for which @xmath196 the map @xmath197 is still an isomorphism .
let @xmath198 be the duality map obtained by the riesz representation theorem .
the defining relation is @xmath199 for all @xmath181 .
it follows from that the operator @xmath198 is the operator associated with the sesquilinear form which is the scalar product in @xmath175 .
note that the sesquilinear form is bounded , with constant @xmath200 , and has coercivity constant which is also equal to @xmath200 .
one deduces from proposition [ p - lions ] that for all @xmath186 the equation @xmath201 admits exactly one solution @xmath187 .
consequently , the operator @xmath202 satisfies maximal parabolic regularity on @xmath203 .
therefore we can apply corollary [ c - extrap ] with @xmath204 .
it follows for the operator family @xmath205 in proposition [ p - lions ] that there is an open interval @xmath206 such that the map in is still an isomorphism for all @xmath161 . using lemma [ lemma - t - intgeropo ] and the sneiberg theorem we also prove a quantitative result on @xmath207 .
the main theorem of this section is the following .
[ thm - e - hilbert ] adopt the notation and assumptions of propositions [ p - lions ] .
let @xmath208 , @xmath209 and define @xmath75 such that @xmath210 .
set @xmath211 let @xmath85 .
let @xmath50 be such that @xmath212 .
then one has the following .
[ thm-0001 - 1 ] if @xmath213 then @xmath214 satisfies maximal @xmath215-regularity .
[ thm-0001 - 2 ] if @xmath216 then @xmath217 where the norm on @xmath218 is defined using the operator @xmath202 . for the proof ,
we first show that injectivity is preserved .
[ l - injective ] adopt the notation and assumptions of propositions [ p - lions ] .
then the map @xmath219 is injective for all @xmath50 . for all @xmath25
the operator @xmath156 is accretive on @xmath203 .
then the claim follows from ( * ? ? ?
* proposition 3.2 ) . ` [ thm-0001 - 1 ] ' .
we apply theorem [ t - sneib ] of sneiberg with the spaces @xmath220 , @xmath221 , @xmath222 and @xmath223
. then @xmath224_\theta = { { \rm mr}}^2_0(j;v , v^*)\,\widetilde{\;}\;$ ] and @xmath225_\theta = l^2(j;v^*)$ ] , with equality of norms by lemma [ lemma - t - intgeropo ] and proposition [ p - interlpp ] .
moreover , @xmath226 is an isomorphism from @xmath227 onto @xmath228 by proposition [ p - lions ] .
we first estimate @xmath229_\theta \to [ f_1,f_2]_\theta}$ ] .
let @xmath230_\theta$ ] .
then @xmath231_\theta } & = & \|u\|_{{{\rm mr}}^2_0(j;v , v^*)\,\widetilde{\ ; } } \leq \|u\|_{{{\rm mr}}^2_0(j;v , v^ * ) } \leq \big ( \frac{1+c_\bullet + c^\bullet } { c_\bullet } \big ) \|\big ( \frac{\partial } { \partial t } + { { \cal b}}(\cdot ) \big ) u\| _ { l^2(j;v^ * ) } , \end{aligned}\ ] ] where we used ( [ ed - tilde;2 ] ) and ( [ ep - lions;10 ] ) in the two inequalities .
so @xmath232 . next we estimate @xmath233 which is the second input for the calculation of the admissible interpolation parameters in theorem [ t - sneib ] .
let @xmath234 and @xmath235 .
then @xmath236 where we used ( [ ed - tilde;1 ] ) in the second step . hence @xmath237 the operator @xmath219 is obviously continuous , and , by lemma [ l - injective ] , it is injective .
moreover , it is surjective by theorem [ t - sneib][p - sneib-1 ] for the claimed interpolation parameters
. then theorem [ thm - e - hilbert ] follows from the open mapping theorem . `
[ thm-0001 - 2 ] ' .
this is proved analogously by using theorem [ t - sneib][p - sneib-2 ] .
[ r - pich ] it would be very interesting to get upper and lower bounds on the exponents @xmath11 , dependent of the constants @xmath238 and @xmath239 .
the difficulty is to get explicit estimates of @xmath240 dependent on @xmath11 . for results related to this problem
, we refer to @xcite .
we continue with some motivation for theorem [ thm - e - hilbert ] . in proposition [ p - lions ] one
has @xmath186 and therefore @xmath241 .
so @xmath55 is continuous from @xmath24 into @xmath176 . in many settings
, it turns out that @xmath242 for some @xmath243 , because in many applications like spatial discretization , one is confronted with step functions . then theorem [ thm - e - hilbert ]
gives that @xmath244 .
we show in the next proposition that then the function @xmath245 is hlder continuous . moreover , the orbits are relatively compact in @xmath176 , if the embedding @xmath246 is compact . this additional regularity can be essential for studying related semi- and quasilinear problems ( cf . section [ snonaut00 ] ) , large - time asymptotic behaviour ( cf .
@xcite ) , or for optimal control problems with tracking type objective functional ( cf . @xcite ) .
[ pnonaut410 ] let @xmath175 and @xmath176 be hilbert spaces with @xmath175 densely and continuously embedded into @xmath176 .
then one has the following .
[ pnonaut410 - 1 ] if @xmath50 and @xmath247 , then @xmath248 .
[ pnonaut410 - 2 ] if @xmath50 and @xmath247 , then @xmath249 , where @xmath250 .
[ pnonaut410 - 3 ] if @xmath251 , then @xmath252 for all @xmath253 .
[ pnonaut410 - 4 ] if @xmath251 and @xmath175 embeds compactly into @xmath203 , then @xmath254 and the embedding is compact . `
[ pnonaut410 - 1 ] ' .
see ( * ? ? ?
* theorem iii.4.10.2 ) . `
[ pnonaut410 - 2 ] ' .
see ( * ? ? ?
* theorem 3 ) , or ( * ? ? ?
* lemma 2.11(b ) ) for an elementary proof . ` [ pnonaut410 - 3 ] ' . set @xmath255
. then @xmath256 .
it follows from statement [ pnonaut410 - 2 ] that @xmath249 .
moreover , @xmath257_\frac{1}{2 } = h$ ] . hence @xmath252 . ` [ pnonaut410 - 4 ] ' .
choose @xmath253 .
set @xmath255 as in the proof of statement [ pnonaut410 - 3 ] .
since @xmath175 embeds compactly into @xmath203 , the embedding @xmath258 is compact by ( * ? ? ?
* section 3.8 ) .
then the statement follows by the ascoli theorem for vector - valued functions , see ( * ? ? ?
* section iii.3 ) .
in the sequel we will apply the above results for the derivation of regularity for non - autonomous parabolic differential operators on sobolev spaces .
we first introduce the underlying elliptic setting . throughout the rest of this paper
we fix a bounded open set @xmath259 , where @xmath260 .
let @xmath47 be a closed subset of the boundary @xmath261 ( to be understood as the dirichlet boundary part ) .
regarding our geometric setting , we suppose the following general conditions .
[ assu - general ] [ assu - general : i ] for every @xmath262 there exists an open neighbourhood @xmath263 of @xmath264 and a bi - lipschitz map @xmath265 from @xmath263 onto the cube @xmath266 , such that the following three conditions are satisfied : @xmath267 [ assu - general : ii ] the set @xmath47 satisfies the _ ahlfors david condition _
, that is there are @xmath268 and @xmath269 , such that @xmath270 for all @xmath271 and @xmath272}$ ] , where @xmath273 denotes the @xmath274-dimensional hausdorff measure and @xmath275 denotes the ball with centre @xmath264 and radius @xmath11 . [ assu - general : iii ] the set @xmath35 is a @xmath276-set in the sense of jonsson wallin ( * ? ? ?
* chapter ii ) .
this condition is , however , * not * needed if the coefficient function @xmath277 or @xmath278 is hermitian , which we introduce below .
that is , if @xmath279 is symmetric or @xmath280 is symmetric for all @xmath281 and @xmath25 , then we do not need that @xmath35 is a @xmath276-set .
[ r - surfmeas ] [ r - surfmeas-3 ] we emphasize that the cases @xmath282 or @xmath283 are allowed .
[ r - surfmeas-1 ] condition means that @xmath47 is a @xmath274-set in the sense of jonsson wallin ( * ? ? ?
* chapter ii ) .
[ r - surfmeas : ii ] on the set @xmath284 the measure @xmath273 is equal to the surface measure @xmath285 , which can be constructed via the bi - lipschitz charts @xmath265 around these boundary points , see ( * ? ? ?
* section 3.3.4 c ) or ( * ? ? ?
* section 3 ) .
in particular , assures that @xmath286 if @xmath287 and @xmath288 .
[ r - surfmeas-5 ] condition [ assu - general : iii ] excludes _ outward _ cusps , but _
inward _ cusps are allowed .
this condition is only used in proposition [ p-3 ] .
[ d-1 ] for all @xmath289 we define the space @xmath290 as the completion of @xmath291 with respect to the ( standard ) norm @xmath292 , where @xmath293 . here
, the dual is to be understood with respect to the extended @xmath294 scalar product , or , in other words , @xmath295 is the space of continuous antilinear functionals on @xmath290 . since the domain @xmath35 is kept fixed , we omit the symbol ` @xmath35 ' in the sequel if no confusion is possible .
for example , we write @xmath296 instead of @xmath290 . the spaces @xmath296 and the spaces @xmath297 interpolate with respect to the complex interpolation functor .
[ l-1 ] adopt assumption [ assu - general ] .
let @xmath298 and @xmath75 . then @xmath299_\theta & = & w^{1,q}_{\mathfrak d}\quad \mbox{and } \\ { } [ w^{-1,q_1}_{\mathfrak d},w^{-1,q_2}_{\mathfrak d}]_\theta & = & w^{-1,q}_{\mathfrak d } , \end{aligned}\ ] ] where @xmath300 .
see @xcite theorem 3.3 and corollary 3.4 .
[ r - fortsetz ] by ( * ? ? ?
* lemma 3.2 ) there exists an extension operator @xmath301 , which is universal in @xmath289 .
therefore one has the usual sobolev embeddings available , including compactness .
we now turn to the definition of the elliptic divergence form operators that will be investigated . for all @xmath302
we denote by @xmath303 the set of all measurable @xmath304 such that @xmath305 for all @xmath306 and almost all @xmath281
. moreover , define @xmath307 the set of all elliptic coefficient functions .
[ d-2 ] for all @xmath308 and @xmath309 define the operator @xmath310 by @xmath311 where @xmath312 and @xmath313 .
then @xmath314 for all @xmath308 by hlder s inequality . moreover , for all @xmath315 define the operator @xmath316 in @xmath297 by @xmath317 and @xmath318 .
define the sesquilinear form @xmath319 by @xmath320 then @xmath321 is coercive and continuous .
let @xmath322 be the @xmath323-sectorial operator in @xmath294 associated to @xmath321 .
so @xmath324 and @xmath325 .
let @xmath326 be the semigroup generated by @xmath327 .
then @xmath326 extends consistently to a contraction semigroup @xmath328 on @xmath329 for all @xmath289 .
moreover , @xmath328 is a holomorphic semigroup on @xmath329 for all @xmath289 by @xcite theorem 3.1 . for all @xmath289
we denote by @xmath330 the generator of @xmath328 . clearly if @xmath331 and @xmath332 , then @xmath333 .
thus the graph of @xmath334 is an extension of the graph of @xmath335 . as a consequence , @xmath336 for all @xmath315 .
similarly , the graph of @xmath337 is an extension of the graph of @xmath338 if @xmath339 and @xmath332 .
in this section we prove that there exists a @xmath340 such that the operator @xmath341 or @xmath342 satisfies maximal parabolic regularity on the space @xmath296 for all @xmath343 .
a key tool is a recent solution of the kato square root problem . in order to determine
@xmath344 we need a definition .
[ d - isoindex ] let @xmath309 .
we call a number @xmath308 an * isomorphism index * for the coefficient function @xmath277 if @xmath345 is a topological isomorphism .
we denote by @xmath346 the set of isomorphism indices for @xmath277 .
although the set @xmath346 also depends on the set @xmath47 , we suppress the dependence of @xmath47 in the notation . if @xmath308 and @xmath309 , then by duality one obviously has @xmath347 if and only if @xmath348 .
this allows to concentrate to all @xmath315 .
[ lnonauto302 ] let @xmath309 and @xmath349 .
then one has the following .
[ lnonauto302 - 1 ] @xmath347 if and only if the operator @xmath341 is surjective .
[ lnonauto302 - 2 ] if @xmath347 , then @xmath350 . `
[ lnonauto302 - 1 ] ' .
. then @xmath352 and @xmath353 for all @xmath354 . by continuity and density
the latter is then also valid for all @xmath355 , in particular for @xmath356 .
since @xmath277 is elliptic one deduces that @xmath357 .
so the operator @xmath341 is injective for all @xmath315 . `
[ lnonauto302 - 2 ] ' .
let @xmath358 . then @xmath359 and @xmath360 .
since @xmath341 is surjective , there exists a @xmath361 such that @xmath362
. then @xmath355 and @xmath363 .
so @xmath364 .
since @xmath365 is injective , one deduces that @xmath366 .
so @xmath367 .
the reverse inclusion is trivial .
the next proposition states that the set @xmath346 is always non - empty and open .
[ p-2 ] adopt assumption [ assu - general ] .
then for all @xmath309 the set @xmath346 is an open interval which contains 2 .
moreover , for all @xmath302 there are @xmath368 such that @xmath369 for all @xmath370 , and , in addition , @xmath371 if follows from lemma [ l-1 ] that @xmath346 is connected .
moreover , @xmath372 by the lax milgram theorem . for the other assertions , see ( * ? ? ?
* theorem 5.6 and remark 5.7 ) .
assumption [ assu - general ] implies that the kato problem for the operator @xmath338 with real measurable coefficients and boundary conditions is solved on @xmath329 for all @xmath373 $ ] .
[ p-3 ] adopt assumption [ assu - general ] .
let @xmath309 and @xmath373 $ ]
. then @xmath374 .
moreover , the operator @xmath375 is a topological isomorphism from @xmath296 onto @xmath329 .
hence its adjoint map @xmath376 is a topological isomorphism from @xmath72 onto @xmath377 for all @xmath378 .
the case @xmath379 is proved in ( * ? ? ?
* main theorem 4.1 ) .
the general case is in ( * ? ? ?
* theorem 5.1 ) .
let @xmath309 and @xmath308 .
if @xmath374 , then it follows from the closed graph theorem that the operator @xmath375 is a topological isomorphism from @xmath296 onto @xmath329 . as a consequence we can split the problem whether @xmath341 is an isomorphism from @xmath296 onto @xmath297 into two parts : from @xmath296 into @xmath329 and from @xmath329 into @xmath297 . [ thmc-001 ] adopt assumption [ assu - general ] .
[ thmc-001 - 1 ] let @xmath315 and @xmath309 .
then @xmath347 if and only if @xmath374 .
[ thmc-001 - 1.5 ] let @xmath309 and @xmath347 .
then @xmath380 and @xmath381 for all @xmath382 .
[ thmc-001 - 2 ] for all @xmath302 there exists an @xmath383 such that @xmath374 for all @xmath384 and @xmath370 . `
[ thmc-001 - 1 ] ' .
suppose that @xmath374 .
then @xmath375 is an isomorphism from @xmath296 onto @xmath329 .
write @xmath385 .
then it follows from proposition [ p-3 ] that @xmath386 and the operator @xmath387 is a topological isomorphism from @xmath388 onto @xmath72 .
let @xmath312 .
we first show that @xmath389 for all @xmath390 .
let @xmath391 .
then @xmath392 and @xmath393 so ( [ ethmc-001;1 ] ) is valid for all @xmath391 .
clearly @xmath394 is continuous from @xmath395 into @xmath180 .
also @xmath396 is continuous from @xmath388 into @xmath180 and hence from @xmath395 into @xmath180 .
since @xmath397 is dense in @xmath398 , and hence in @xmath395 , it follows by continuity that ( [ ethmc-001;1 ] ) is valid for all @xmath399 .
let @xmath400 .
by the last part of proposition [ p-3 ] there exists a @xmath401 such that @xmath402 for all @xmath390 . by assumption
there exists a @xmath312 such that @xmath403 .
then @xmath404 for all @xmath390 , where we used ( [ ethmc-001;1 ] ) in the last step .
so @xmath405 and @xmath406 is surjective
. therefore @xmath347 by lemma [ lnonauto302][lnonauto302 - 1 ] .
next let @xmath347 .
we shall show that @xmath407 .
. then @xmath408 and @xmath409 .
by the last part of proposition [ p-3 ] there exists a @xmath401 such that @xmath410 for all @xmath390 , where @xmath385 .
let @xmath411 .
then @xmath412 since @xmath413 is a core for @xmath337 one deduces that @xmath414 for all @xmath415 .
hence @xmath416 for all @xmath417 .
this implies that @xmath418 .
conversely , let @xmath419
. then @xmath420 . by the last part of proposition [ p-3 ] there
exists a @xmath400 such that @xmath421 for all @xmath390 . since @xmath347
, the operator @xmath341 is surjective .
hence there exists a @xmath422 such that @xmath423 .
now let @xmath411 .
then @xmath424 since @xmath425 is surjective , it follows that @xmath426 . `
[ thmc-001 - 1.5 ] ' .
again write @xmath385 .
first suppose that @xmath427 .
. then @xmath428 by statement [ thmc-001 - 1 ] .
moreover , @xmath429 . if @xmath430 then @xmath431 and ( [ ethmc-001;1 ] ) gives @xmath432 hence @xmath433 .
conversely , let @xmath312 and suppose that @xmath434 .
write @xmath435 .
then @xmath436 for all @xmath437 , where we used ( [ ethmc-001;1 ] ) .
it follows that @xmath438 .
hence @xmath439 . now suppose that @xmath440 .
let @xmath382 .
then @xmath428 by proposition [ p-3 ] .
if @xmath415 then by the above @xmath390 and @xmath441 .
so @xmath442 and @xmath443 since @xmath398 is dense in @xmath386 , one deduces that ( [ ethmc-001;2 ] ) is valid for all @xmath390 .
so @xmath433 .
conversely , suppose that @xmath312 and @xmath434 .
let @xmath415 .
then again by the above @xmath444 so @xmath445 and the proof of statement [ thmc-001 - 1.5 ] is complete . `
[ thmc-001 - 2 ] ' .
this follows from statement [ thmc-001 - 1 ] and proposition [ p-2 ] .
we next present a few illustrative examples with explicit subsets of the set @xmath346 .
note that the requirements on the geometry of @xmath35 and the dirichlet boundary part @xmath47 , as well as on the coefficient function @xmath277 is much higher in the examples than in our general assumptions .
[ xnonaut317.3 ] assume that @xmath35 is a @xmath446-domain and that @xmath447 or @xmath448 ( pure dirichlet or pure neumann condition ) .
if @xmath309 is uniformly continuous on @xmath35 , then @xmath449 by ( * ? ? ?
* section 15 ) or @xcite .
the conclusion remains true , if there is a @xmath446-subdomain @xmath450 with positive distance to the boundary , such that @xmath451 and @xmath452 are uniformly continuous , see ( * ? ? ?
* theorem 1.1 and remark 3.15 ) .
[ xnonaut317.5 ] assume that @xmath35 is a lipschitz graph - domain ( see ( * ? ? ?
* definition 1.2.1.1 ) ) .
there are equivalent terminologies for this notion : strong lipschitz domain in ( * ? ? ?
* section 1.1.8 ) and @xmath35 possesses the uniform cone property in ( * ? ? ?
* section 1.2.2 ) .
suppose that @xmath309 takes symmetric matrices as values .
then , under the same continuity properties for @xmath277 as in example [ xnonaut317.3 ] ( both cases ) , the ( open ) set @xmath346 contains the interval @xmath453 $ ] both in the pure dirichlet case ( that is @xmath447 ) , and in the pure neumann case ( that is @xmath448 ) , see @xcite .
moreover , one can not replace @xmath454 by a larger number , independent of @xmath35 and @xmath277 . for the pure dirichlet laplacian this result
was already proved in ( * ? ? ?
* theorem 1.1(c ) and theorem 1.2(a ) ) , and for the pure neumann laplacian in @xcite .
[ xnonaut317.8 ] in @xcite there are given a huge variety of domains @xmath455 , dirichlet boundary parts @xmath47 and ( possibly discontinuous even up to the boundary ) elliptic coefficient functions @xmath277 , such that @xmath346 contains the interval @xmath453 $ ] .
in particular , it is allowed that @xmath456 is not empty , i.e. the dirichlet boundary part meets the neumann part . for all
@xmath308 we consider the operator @xmath335 as a densely defined operator in @xmath297 with domain @xmath388 .
let @xmath309 and @xmath457 .
write @xmath385 .
then it follows from proposition [ p-3 ] and theorem [ thmc-001][thmc-001 - 1 ] that @xmath458 is a topological isomorphism .
let @xmath459 be the adjoint of the operator .
then @xmath460 is an isomorphism too .
we use the isomorphism @xmath460 to transfer the @xmath461-semigroup @xmath328 on @xmath329 to a @xmath461-semigroup @xmath462 on @xmath297 .
explicitly , for all @xmath463 define @xmath464 by @xmath465 then @xmath462 is a @xmath461-semigroup on @xmath297 .
clearly @xmath460 is an extension of the operator @xmath375 and hence @xmath466 is an extension of @xmath467 .
since @xmath467 and @xmath468 commute for all @xmath469 , it follows that @xmath470 is an extension of @xmath468 for all @xmath469 .
we denote the generator of @xmath462 by @xmath471 .
obviously @xmath472 is consistent with @xmath473 for all @xmath474 .
hence the graph of @xmath475 is an extension of the graph of @xmath476 and @xmath477 if @xmath478 .
[ lnonaut304 ] adopt assumption [ assu - general ] .
let @xmath277 be in @xmath479 .
[ lnonaut304 - 1 ] if @xmath457 , then @xmath480 .
[ lnonaut304 - 2 ] if @xmath347 , then @xmath481 .
[ lnonaut304 - 3 ] if @xmath315 , then @xmath482 . `
[ lnonaut304 - 1 ] ' . by definition of the semigroup
@xmath462 it follows that @xmath483 write @xmath385 . if @xmath484 and @xmath390 , then @xmath485 so @xmath486 for all @xmath484 .
hence @xmath487 this proves statement [ lnonaut304 - 1 ] . `
[ lnonaut304 - 2 ] ' .
. then @xmath488 by statement [ lnonaut304 - 1 ] , theorem [ thmc-001][thmc-001 - 1 ] and proposition [ p-3 ] .
if @xmath484 and @xmath390 , then @xmath489 where the last equality is ( [ ethmc-001;1 ] ) .
so @xmath490 for all @xmath484 .
hence @xmath491 for all @xmath492 . using again that @xmath486 one deduces that @xmath493 for all @xmath492
hence @xmath481 . `
[ lnonaut304 - 3 ] ' .
let @xmath315 .
it follows from ( [ elnonaut304;1 ] ) and statement [ lnonaut304 - 2 ] that @xmath494 so @xmath482 . in ( [ elnonaut304;30 ] )
the topological isomorphism @xmath495 was used to define the @xmath461-semigroup @xmath462 from the @xmath461-semigroup @xmath328 .
it then transfers properties of the generator of @xmath328 to properties of the generator of @xmath462 .
[ tnonaut610 ] adopt assumption [ assu - general ] .
let @xmath309 and @xmath457 .
then the operator @xmath496 satisfies maximal parabolic regularity on the space @xmath297 .
the semigroup @xmath328 is a contraction semigroup , hence the operator @xmath497 has maximal parabolic regularity in the space @xmath329 by lamberton @xcite . since @xmath495 is a topological isomorphism from @xmath329 onto @xmath297 , where @xmath385 , it follows from ( [ elnonaut304;30 ] ) that the operator @xmath496 satisfies maximal parabolic regularity on the space @xmath297 .
note that @xmath481 for all @xmath347 in the next theorem .
the case @xmath315 in the next corollary has been proved before in ( * ? ? ?
* theorem 11.5 ) .
[ cnonaut611 ] adopt assumption [ assu - general ] .
then for all @xmath302 there exists a @xmath498 such that the operator @xmath496 satisfies maximal parabolic regularity on the space @xmath297 for all @xmath499 and @xmath370 .
this follows immediately from proposition [ p-2 ] and theorem [ tnonaut610 ] .
[ r - explain1 ] it is clear that there is an asymmetry in the cases @xmath315 and @xmath373 $ ] . in the first case , maximal parabolic regularity holds for the operators @xmath500 on @xmath297 by theorem [ tnonaut610 ] and lemma [ lnonaut304][lnonaut304 - 3 ] , even if the domain of this operator is unknown . on the contrary , in the case @xmath501 , we can only prove maximal parabolic regularity for the operator @xmath341 in @xmath297 if @xmath347 .
this is a severe restriction on @xmath32 , see examples [ xnonaut317.3][xnonaut317.8 ] .
it is an open problem whether in corollary [ cnonaut611 ] maximal parabolic regularity for @xmath341 is valid on @xmath297 for all @xmath502 .
we next consider coefficient functions which also may depend on time .
let @xmath503 be a function .
we frequently write @xmath504 for all @xmath25 . note that @xmath505 for all @xmath25 .
we say that * @xmath277 is @xmath506-measurable * if the map @xmath46 is measurable as a map from @xmath24 into @xmath507 .
in the main result of this section we require measurability of the coefficient function only in the space @xmath507 .
this allows that @xmath27 is discontinuous in the space variable for each @xmath25 .
note that the set of point in @xmath35 where @xmath27 is discontinuous may depend on @xmath23 .
in general the map @xmath46 from @xmath24 into @xmath508 is discontinuous at _ every _ time point @xmath23 and therefore it can not be measurable .
an example is mentioned in the introduction and it will be considered in more detail in section [ snonaut8 ] .
[ lnonaut321.5 ] adopt assumption [ assu - general ] .
let @xmath509 and @xmath510 be an @xmath506-measurable map .
let @xmath511 .
then one has the following .
[ lnonaut321.5 - 1 ] the map @xmath512 is ( strongly ) measurable from @xmath24 into @xmath297 for all @xmath312 .
[ lnonaut321.5 - 2 ] the map @xmath513 is a bounded linear map from @xmath514 into @xmath13 with norm at most @xmath515
. let @xmath516 .
then the map @xmath517 is continuous from @xmath518 into @xmath180 .
since @xmath277 is @xmath506-measurable , also the map @xmath519 from @xmath24 into @xmath180 is well defined , bounded and measurable . since @xmath520 is dense in @xmath296 and @xmath521 and @xmath277
is bounded , the map ( [ lnonaut321.5;1 ] ) is measurable for all @xmath312 and @xmath313 .
therefore one obtains the weak measurability of the map @xmath522 for all @xmath312 , which implies also the strong measurability , since the space @xmath297 is separable .
this proves the first statement .
the second one is easy .
the first main result of this section is as follows . in order to get good estimates ,
we use again the normed spaces @xmath523 introduced in definition [ d - tilde ] .
[ tnonaut701 ] let @xmath524 with @xmath525 .
let @xmath526 and put @xmath527 where @xmath528 is the duality map .
define @xmath529 then for every @xmath506-measurable @xmath530 and @xmath531 , the family @xmath532 has maximal @xmath533-regularity and @xmath534 where the norm on @xmath523 is defined using the operator @xmath202 . we wish to apply theorem [ thm - e - hilbert ] .
let @xmath535 .
then @xmath536 .
note that @xmath537 if @xmath538 .
let @xmath530 be an @xmath506-measurable map .
for all @xmath25 define @xmath539 by @xmath540 = { \gothic{s}}_{\mu(t)}[\psi,\varphi ] + ( \psi,\varphi)_{l^2 } .\ ] ] then @xmath541 $ ] is measurable from @xmath24 into @xmath180 for all @xmath542 .
moreover , @xmath182 \ge c_\bullet\|\psi\|^2_v$ ] and @xmath183| \le c^\bullet \ , \|\psi\|_v \ , \|\varphi\|_v$ ] for all @xmath184 and @xmath25 . if @xmath25 , then @xmath543 is the operator induced by the sesquilinear form @xmath544 .
all the assumptions of theorem [ thm - e - hilbert ] are satisfied .
if @xmath545 , then it follows from theorem [ thm - e - hilbert][thm-0001 - 2 ] that the isomorphism property is preserved . then the assertion follows by using the identity @xmath546 .
the second main result of this section is that non - autonomous maximal @xmath13-regularity extrapolates in both temporal and spatial integrability scales , given by @xmath11 and @xmath32 .
again , quantitative estimates as in below , are based on the constants @xmath547 in , corresponding to a suitable autonomous reference operator @xmath548 .
[ t - main ] suppose assumption [ assu - general ] is satisfied .
let @xmath302 .
then there are open intervals @xmath549 with @xmath550 and @xmath551 such that for all @xmath552 , @xmath553 and @xmath506-measurable @xmath530 the family @xmath554 has maximal parabolic @xmath13-regularity .
so @xmath555 is a topological isomorphism .
define @xmath556 by @xmath557 , the identity matrix , for all @xmath281 . then @xmath558 and @xmath559 , the minus laplacian .
let @xmath528 be the duality mapping .
then @xmath560 for all @xmath561 .
so @xmath562 .
if follows from proposition [ p-2 ] , lemma [ lnonaut304][lnonaut304 - 2 ] and theorem [ tnonaut610 ] that there exists a @xmath563 such that @xmath564 is an isomorphism and the operator @xmath565 satisfies maximal parabolic regularity on the space @xmath297 for all @xmath566 . by proposition [ p-2 ] ,
there is a @xmath567 $ ] such that @xmath568 \subset { \mathfrak i}_{\tilde \mu}$ ] for all @xmath569 . for all @xmath570 $ ]
, we choose @xmath571 as the autonomous reference operator for the spaces @xmath572 in definition [ d - tilde ] .
note that in case @xmath379 the reference operator is @xmath573 , which was used in theorem [ tnonaut701 ] . for all @xmath526 and @xmath574 with @xmath575
we obtain by lemma [ lemma - t - intgeropo ] that @xmath576_\alpha = { { \rm mr}}^r_0(j;w^{1,q}_{\mathfrak d},w^{-1,q}_{\mathfrak d})\,\widetilde{\;}\ ] ] with equality of norms . moreover , for each @xmath570 $ ] and @xmath50 the norms on @xmath514 and @xmath577 are equivalent .
hence it suffices to prove the theorem with @xmath514 replaced by @xmath572 . by theorem [ tnonaut701 ]
, there is an @xmath578 such that for all @xmath506-measurable @xmath530 and @xmath579 $ ] the map @xmath580 is a topological isomorphism with @xmath581 these will be the important inverse bounds to apply theorem [ t - sneib ] .
next , let @xmath582 .
we need a suitable bound on the operator norms @xmath583 uniformly in @xmath584 $ ] .
since both the spaces @xmath572 and @xmath13 form exact complex interpolation scales in @xmath11 by and proposition [ p - interlpp ] , it follows by interpolation that @xmath585 now let @xmath586 .
then it follows from lemma [ lnonaut321.5][lnonaut321.5 - 2 ] and ( [ ed - tilde;1 ] ) that @xmath587 for all @xmath588 .
so @xmath589 . set @xmath590 then we proved that @xmath591 for all @xmath579 $ ] , @xmath592 and @xmath506-measurable @xmath530 .
let @xmath593 finally , let @xmath530 be @xmath506-measurable .
let @xmath579 $ ] and @xmath594 $ ] .
then there exists a @xmath595 $ ] such that @xmath596 .
note that @xmath597 with @xmath538 .
we apply theorem [ t - sneib ] with @xmath598 , @xmath599 , @xmath600 , @xmath601 and @xmath538 .
note that we have the estimates ( [ et - main;10 ] ) and ( [ beta - est ] ) .
since @xmath602 one deduces from theorem [ t - sneib ] that @xmath603 is a topological isomorphism . this completes the proof of theorem [ t - main ] .
in this section we are interested in quasilinear , non - autonomous equations of the form @xmath604 the main result is the following . [ tquasilin ]
let @xmath259 be a bounded open set and @xmath605 be closed .
suppose assumption [ assu - general ] is satisfied .
let @xmath302 and @xmath530 an @xmath506-measurable map .
let @xmath606 with @xmath607 .
let @xmath608 $ ] be a continuous function .
then there exists an @xmath208 such that for all @xmath609 and @xmath610 there exists a @xmath611 such that @xmath612 in @xmath613 for almost every @xmath25 . by theorem [ tnonaut701 ]
there exist @xmath208 and @xmath614 such that for every @xmath506-measurable @xmath615 the family @xmath616 has maximal @xmath533-regularity and @xmath617 for all @xmath618 $ ] .
now let @xmath619 $ ] , @xmath610 and @xmath530 be an @xmath506-measurable map .
we wish to define a map @xmath620 .
let @xmath621 .
then @xmath622 for almost every @xmath25 and @xmath623 is @xmath506-measurable .
hence there exists a unique @xmath624 such that @xmath625 for almost every @xmath25
. then @xmath626 by proposition [ pnonaut410][pnonaut410 - 4 ] .
define @xmath627 .
then @xmath628 is relatively compact in @xmath629 by proposition [ pnonaut410][pnonaut410 - 4 ] .
we next show that @xmath630 is continuous .
then the theorem follows from schauder s fixed point theorem .
let @xmath631 and suppose that @xmath632 in @xmath629 .
for all @xmath633 let @xmath634 and @xmath635 .
then @xmath636 , @xmath637 and @xmath638 for almost every @xmath25 and all @xmath633 .
subtracting gives @xmath639 for almost every @xmath25 and all @xmath633 .
since @xmath640 in @xmath629 , also @xmath640 in @xmath641 . hence passing to a subsequence , if necessary
, we may assume that @xmath642 for almost every @xmath643 .
for all @xmath633 define @xmath644 by @xmath645 we shall show that @xmath646 weakly in @xmath533 . let @xmath647
. then @xmath648 obviously @xmath649 for all @xmath633 .
hence @xmath650 by the lebesgue dominated convergence theorem .
so @xmath646 weakly in @xmath533 . but @xmath651 for almost every @xmath25 and all @xmath633 by ( [ etquasilin-2 ] ) .
also @xmath652 for all @xmath633 . hence by maximal parabolic regularity @xmath653 weakly in @xmath654 .
in addition the embedding of @xmath654 into @xmath629 is compact by proposition [ pnonaut410][pnonaut410 - 4 ] .
so @xmath653 in @xmath629 and the continuity of @xmath630 follows .
[ c - morereg ] if the right hand side @xmath51 in belongs to a space @xmath13 and @xmath655 are sufficiently close to @xmath12 , then every solution @xmath55 provided by the theorem belongs to the space @xmath656 .
the coefficient function @xmath657 satisfies the assumptions of theorem [ t - main ] .
it is unclear whether has a unique solution .
let @xmath35 and @xmath47 satisfy assumption [ assu - general ] .
let @xmath658 be an open non - empty set such that @xmath659 . define a reference coefficient function @xmath660 by @xmath661 2
\ , i_d & \mbox{if } x \in \omega \setminus { \omega_0 } , \end{array } \right.\ ] ] where @xmath662 denotes the identity matrix in @xmath663 . for all @xmath664 let @xmath665 be a map and set @xmath666 .
suppose that @xmath46 is @xmath506-measurable .
then by theorem [ t - main ] the operator family @xmath667 satisfies maximal parabolic @xmath668-regularity for all @xmath669 sufficiently close to @xmath12 .
more specifically , consider the case in which @xmath670 is an open subset of @xmath35 for all @xmath664 such that @xmath671 for all @xmath672 with @xmath673 .
then the map @xmath674 from @xmath24 into @xmath675 is discontinuous at every point @xmath664 and it is straightforward to show that also the map @xmath676 from @xmath24 into @xmath677 is discontinuous .
we remark that the analysis for this problem is known to be complicated already in case of elliptic equations , see the discussion in @xcite for relatively simple geometries of interfaces .
moreover , it represents a challenge also in numerics , see e.g. @xcite . in many interesting cases ,
the movement of the subdomain @xmath678 is not determined by an ` outer ' law , but may depend on the underlying physical / chemical process itself .
mathematically , this leads to a free boundary problem where for example @xmath679 depends on the solution @xmath55 .
particular ( simple ) cases may then be covered by theorem [ tquasilin ] to obtain existence and regularity of a solution .
[ r - real ] it is possible to carry our results over to real spaces : in case of the real space @xmath680 one identifies its dual with the elements of @xmath681 which take real values for real functions from @xmath682 .
then one applies the ` complex ' result .
this is enabled by the fact that , in case of real coefficients , the corresponding operators map the real subspace onto the ` real ' subspace of the image .
we expect that our abstract results in section [ snonaut4 ] have further applications in the field of maximal parabolic regularity for non - autonomous parabolic equations .
for example , one could investigate non - autonomous problems in the @xmath683-setting , cf .
* section 5 ) , @xcite and @xcite .
maximal parabolic regularity for autonomous elliptic , second - order divergence - form operators @xmath684 on @xmath33 , with @xmath685 , can be shown under assumption [ assu - general][assu - general : i ] . in this case , however , it is very difficult to determine the exact domain @xmath686 of an operator @xmath687 , if the coefficient function is spatially discontinuous .
the condition @xmath688 in our results then generically excludes settings like the one in section [ snonaut8 ] . at the same time , recent optimal results in @xcite on varying domains with @xmath686 require some continuity in time and regularity in space which also do not cover this setting . in particular , we highlight that in @xcite it is shown that the extrapolation for lions result , proposition [ p - lions ] , is impossible in an @xmath72-setting .
the abstract results in section [ snonaut4 ] and applications to non - autonomous forms in section [ snonaut3 ] naturally include systems of equations . for the more specific setting in sections [ snonaut5][snonaut7 ] ,
one has the required elliptic @xmath689-regularity for systems ( see ( * ? ? ?
* section 6 ) or ( * ? ? ?
* section 7 ) ) , but presently the corresponding maximal parabolic regularity results are an open problem . height 2pt depth -1.6pt width 23pt , maximal regularity and quasilinear parabolic boundary value problems . in
_ recent advances in elliptic and parabolic problems _ , 117 .
world sci .
hackensack , nj , 2005 . , semigroups and evolution equations : functional calculus , regularity and kernel estimates . in dafermos , c. m. and feireisl , e. , eds . , _ evolutionary equations .
i _ , handb . differ .
north - holland , amsterdam , 2004 .
height 2pt depth -1.6pt width 23pt , maximal parabolic regularity for divergence operators on distribution spaces . in j.
escher et .
, _ parabolic problems , the herbert amann festschrift _ , progress in nonlinear differential equations and their applications 80 .
birkhuser , basel , 2011 , 313342 . | we consider linear inhomogeneous non - autonomous parabolic problems associated to sesquilinear forms , with discontinuous dependence of time .
we show that for these problems , the property of maximal parabolic regularity can be extrapolated to time integrability exponents @xmath0 .
this allows us to prove maximal parabolic @xmath1-regularity for discontinuous non - autonomous second - order divergence form operators in very general geometric settings and to prove existence results for related quasilinear equations . * on maximal parabolic regularity for non - autonomous parabolic operators * karoline disser , a.f.m .
ter elst , joachim rehberg _ key words and phrases : _ non - autonomous evolution equations , parabolic initial boundary value problems , maximal parabolic regularity , extrapolation of maximal parabolic regularity _ 2010 mathematics subject classification .
_ 35b65 ( primary ) , 47a07 , 35k20 , 35b45 , 46b70 _ acknowledgements .
_ k.d . was supported by the european research council via `` erc-2010-adg no .
267802 ( analysis of multiscale systems driven by functionals ) '' . |
one of the basic assumptions in modern cosmology , the _ cosmological principle _
, is that on large - scale average our universe is spatially homogeneous and isotropic .
the apparent isotropy on large scales is normally explained as a consequence of spatial homogeneity , in turn understood as a natural result of an inflationary " period of the early universe ( see e.g. kolb and turner , 1990 ) .
an alternative approach to explaining the apparent homogeneity is to assume an expanding universe with small and finite space sections with a non - trivial topology ( ellis and schreiber , 1986 ) , the small universe " model .
the small universe " , as its name suggests , should be small enough that we have had time to see the universe around us many times since the decoupling time
. the topology of the spatial sections can be quite complicated ( ellis , 1971 ) ; however , it is possible to obtain small universe models that reproduce a friedmann - lematre model by choosing certain simple geometries .
for example , choosing a rectangular basic cell with sides @xmath5 , @xmath6 and @xmath7 and with opposite faces topologically connected , we obtain a toroidal topology for the small universe known as @xmath8 .
the never - ending repetition of this @xmath8 basic cell should reproduce , at least locally , the friedmann - lematre universe model with zero curvature .
the small universe model has received considerable attention in the past few years , since the topology of the universe is becoming an important problem for cosmologists . from the theoretical point of view , it is possible to have quantum creation of the universe with a nontrivial topology , i.e. , a multiply - connected topology ( zeldovich and starobinsky , 1984 ) . from the observational side , this model has been used to explain observed " periodicity in the distributions of quasars ( fang and sato , 1985 ) and galaxies ( broadhurst et al . , 1990 ) . there are four known approaches for placing lower limits on the cell size of the @xmath8 model .
the first two methods constrain the parameter @xmath9 , an average length scale of the small universe , defined as @xmath10 .
the third and fourth methods constrain the parameter @xmath11 , the ratio between the cell size @xmath12 , here defined as @xmath13 , and radius of the decoupling sphere @xmath14 , where @xmath15 .
the first method constrains @xmath9 assuming that it is larger than any distinguishable structure . using this method , fairall ( 1985 )
suggests that @xmath16mpc .
the second method constrains @xmath9 based on observed " periodicity in quasar redshifts . attempting to identify opposite pairs of quasars ,
fang and liu ( 1988 ) suggested that @xmath17mpc and using quasar redshift periodicity , fang and sato ( 1985 ) suggested @xmath18mpc .
the third and fourth methods constrain @xmath11 using the cmb . with the third method , stevens et al .
( 1993 ) obtain the constraint @xmath19 using the slope information from the 1st year of cobe / dmr data ( smoot et al . , 1992 ) while , with the fourth method , jing and fang ( 1994 ) obtain a best fit @xmath20 using the correlation function from the 2-year cobe / dmr data . as pointed out by zeldovich ( 1973 ) , the power spectrum of density perturbations is continuous ( i.e. , all wave numbers are possible ) if the universe has a euclidean topology , and discrete ( i.e. , only some wave numbers are possible ) if the topology has finite space sections .
many years later these ideas were related with the expected cmb power spectrum ( fang and mo , 1987 ; sokolov , 1993 ; starobinsky , 1993 ) , mainly after the quadrupole component had been detected by cobe / dmr .
our goal is to place new and accurate limits on the cell size of a small universe using the harmonic decomposition technique to obtain the data power spectrum ( grski , 1994 ) and likelihood technique ( bunn and sugiyama , 1994 ) to constrain @xmath11 .
the method that we use to constrain the parameter @xmath11 is quite different from previous work .
the method adopted by stevens et al .
( 1993 ) constrains the cell size based in the power spectrum of the cmb ; they graphically compare the power spectrum of the standard model with the power spectrum expected for the small universe , normalizing to the @xmath21 component .
jing and fang ( 1994 ) adopt a different approach : they constrain the cell size using the correlation function of the cmb and making the approximation that bins of the correlation function are uncorrelated .
our analysis , however , is exact .
we compute the full covariance matrix for all multipole components and use this covariance matrix to make a @xmath22 fit of the power spectrum extracted from the 2 years of cobe / dmr data to the power spectrum expected for a small universe with different cell sizes @xmath12 . for simplicity ,
we limit our calculation to the case of a @xmath8 cubic universe .
we present , in the next sections , a description of the power spectrum expected in a @xmath8 cubic universe model and the likelihood technique used to constrain @xmath11 .
if the density fluctuations are adiabatic and the universe is spatially flat , the sachs - wolfe fluctuations in the cmb are given by ( , ) = - _ * k * e^i*k * ( peebles , 1982 ) , where @xmath23 is a vector with length @xmath24 that is pointed in the direction of observation ( @xmath25 ) , @xmath26 is the hubble constant ( written here as @xmath27 km / s / mpc ) and @xmath28 is the density fluctuation in fourier space with the sum taken over all wave numbers * k*. it is customary to expand the cmb anisotropy in spherical harmonics ( , ) = _ l=0^ _ m =- l^l a_lm y_lm ( ) , where @xmath29 are the spherical harmonic coefficients and @xmath30 is the unit vector in direction @xmath23 .
the coefficients @xmath29 are given by a_lm = -2i^l _ * k * j_l(ky ) y_lm^ * ( ) , where @xmath31 are spherical bessel functions of order @xmath32 .
if we assume that the cmb anisotropy is a gaussian random field , the coefficients @xmath29 are independent gaussian random variables with zero mean and variance | a_lm |^2 = 16_*k * j_l^2(ky ) ( fang and mo , 1987 ; stevens et al . , 1993 ) . assuming a power - law power spectrum with shape @xmath33 , where @xmath34 is the amplitude of scalar perturbations and @xmath35 the spectral index , it is possible to perform the sum in ( [ almvariance ] ) , replacing it by an integral , and to obtain |
a_lm |^2 = c_2 ( see e.g. bond and efsthatiou , 1987 ) . in the literature ,
the average over the canonical ensemble of universes @xmath36 is usually denoted by c_l the rms temperature fluctuation by @xmath37 .
note that in a euclidean topology the sachs - wolfe spectrum @xmath38 is an integral over the power spectrum ; however , in the @xmath8 universe this is not the case . in this model ,
only wave numbers that are harmonics of the cell size are allowed .
we have a discrete @xmath39 spectrum ^2 = _
i=1 ^ 3 ( ) ^2 p^2_i ( sokolov , 1993 ) , where @xmath40 , @xmath41 and @xmath42 are the dimensions of the cell and @xmath43 are integers . for simplicity , assuming @xmath13 and the same power - law power spectrum cited before , eq .
( [ almvariance ] ) can be written as |
|^2 = _ p_x _ p_y _ p_z ( ) ^4-n j_l^2 ( ) , where @xmath44 .
according to ( [ almsu ] ) , the @xmath32th multipole of the cmb temperature is function of the ratio @xmath11 .
this shows that the more multipole components we use in our fit , the stronger our constraints on the cell size will be .
however , we can not use an infinite number of multipole components .
the maximum number of multipole components , @xmath45 , will be limited by two things : the limit where the map is noise dominated and the limit where we can truncate the fourier series without compromising the harmonic decomposition technique ( see grski , 1994 ) .
( [ almsu ] ) , we calculated the expected power spectrum for a @xmath8 universe with different cell sizes @xmath11 from 0.1 to 3.0 , @xmath1 and @xmath46 , where @xmath46 is the limit at which we truncate our data power spectrum . in figure 1
, we plot @xmath47 versus @xmath32 and normalize all values to the last multipole component @xmath48 .
note that for very small cells ( @xmath49 ) , the low order multipoles are suppressed .
the power spectrum for small cells ( as @xmath50 , @xmath51 or @xmath52 ) shows the presence of bumps " that disappear as the cell size increases ( @xmath53 ) .
the power spectrum finally becomes flat for large cell sizes ( @xmath54 ) .
these bumps " can be explained if we remember that only the harmonics of the cell size are allowed to be part of the sum in ( [ almsu ] ) .
when the cell size is small there are fewer modes of resonance , and no modes larger than the cell size appear in the sum in ( [ almsu ] ) . as the cell size increases , the sum approaches an integral and the @xmath8 power spectrum becomes flat .
we restrict our analysis to @xmath1 .
this assumption , however , does not weaken our results , since the @xmath8 model with other @xmath35-values tends to fit the data as poorly as with @xmath1 .
for instance , we obtain the maximum likelihood at the same ratio @xmath11 for @xmath1 and @xmath55 .
this happens because the bumps " , and not the overall slope , are responsible for the disagreement between the model and the data .
each dmr sky map is composed of @xmath56 pixels and each pixel @xmath57 contains a measurement of the sky temperature at position @xmath58 . considering that the temperatures are smoothed by the dmr beam and contaminated with noise ,
the sky temperatures are described by ( ) _ i = _ lm
a_lm b_l y_lm(_i ) + n_i , where @xmath59 is the dmr beam pattern and @xmath60 is the noise in pixel @xmath57 .
we use the values of @xmath59 given by wright et al .
( 1994a ) , which describes the actual beam pattern of the dmr horns , an imperfect gaussian beam .
we model the quantities @xmath60 in ( [ deltatobserved ] ) as gaussian random variables with mean @xmath61 and variance @xmath62 , assuming uncorrelated pixel noise ( lineweaver et al . , 1994 ) .
when we have all sky coverage , the @xmath29 coefficients are given by a_lm = _ 4 ( ) y_lm^ * ( ) d. in the real sky maps , we do not have all sky coverage .
because of the uncertainty in galaxy emission , we are forced to remove all pixels between @xmath63 below and above the galaxy plane .
this cut represents a loss of almost 34% of all sky pixels and destroys the orthogonality of the spherical harmonics . replacing the integral in ( [ almcoeff ] ) by a sum over the number of pixels that remain in the sky map after the galaxy cut , @xmath64
, we define a new set of coefficients by b_lm w _ i=1^n_pix ( ) _ i y_lm^*(_i ) , where the normalization is chosen to be @xmath65 . substituting ( [ deltatobserved ] ) into ( [ blmobserved ] )
, we obtain b_lm = _ l_1m_1 a_l_1m_1 b_l_1 w_ll_1mm_1 + w _
i=1^n_pix n_i y_lm^*(_i ) , with covariance @xmath66 where w_ll_1mm_1 w _
i=1^n_pix y_lm^*(_i ) y_l_1m_1(_i ) .
defining our multipole estimates as c_l^dmr _
m b_lm b_lm^ * , their expectation values are simply c_l^dmr _ m b_lm b_lm^ * and their covariance matrix @xmath67 is given by m_ll _
mm b_lm b_lm^ * ^2 .
the @xmath68 coefficients are not good estimates of the true multipole moments @xmath38 .
however , they are useful for constraining our cosmological parameters . the likelihood and the @xmath22 are , respectively , defined by -2 = ^2 + |*m*| and ^2 ^t * m*^-1 * c * , where @xmath69 and @xmath70 are @xmath45-dimensional row and column vectors with entries @xmath71 and @xmath72 is the covariance matrix as described in ( [ clvariance ] ) with dimensions @xmath45 x @xmath45 . here
@xmath73 denotes the @xmath68-coefficients actually extracted from the data .
because the perturbations depend on an unknown constant @xmath34 , the power spectrum normalization , we have to constrain two parameters at once . in practice , this calculation is done by fixing the ratio @xmath11 and changing the normalization by a small factor .
we multiply the first term on the right side of ( [ blmvariance ] ) by this factor and calculate a new covariance matrix . repeating this procedure for each cell size
, we finally get a likelihood grid that constrains the ratio @xmath11 and the normalization parameter .
in figure 2 , we show the angular power spectrum @xmath73 extracted from the data .
we use a 2-year combined 53 plus 90 ghz map , with galaxy cut of @xmath63 , monopole and dipole removed .
we plot @xmath74 versus @xmath32 from @xmath75 to @xmath48 , with bias ( @xmath76 ) removed and error bars given by the diagonal terms of the covariance matrix @xmath67 . in computing the bias and error bars ,
we assume eq .
( [ almpadrao ] ) with @xmath1 .
the shape of this power spectrum and its multipole values are consistent with values reported by wright et al .
( 1994b ) , and for @xmath77 the power spectrum is basically dominated by noise .
we computed the likelihood function @xmath78 , using it to constrain the ratio @xmath11 and the normalization @xmath79 , where @xmath79 is the rms variance at @xmath80 .
for the data set described above , we found the maximum likelihood at @xmath81 . in figure 3 , we plot the likelihood function @xmath78 .
notice that the likelihoods can not be normalized because they do not converge to zero for very large cell sizes , i.e. , the volume under the likelihood function is infinite .
since the likelihoods are not zero for very large cell sizes , we could naively consider that the probability of the universe being small is essentially zero .
however , this conclusion is clearly exaggerated and based on the fact that we multiplied our likelihoods by a uniform prior , and there is nothing special about adopting a uniform prior . in order to obtain rigorous confidence limits for our analysis
, we replace the maximum likelihood fit by a minimum @xmath22 fit .
we compute the chi - squared function @xmath82 and use it to constrain the ratio @xmath11 and the normalization @xmath79 . in figure 4 , we plot the probability that the @xmath8 model is consistent with the data as a function of the ratio @xmath11 and the normalization @xmath79 ( bottom ) .
confidence limits of 68% , 95% and 99.7% are shown in the contour plot ( top ) .
we found the highest consistency probability ( minimum @xmath22 ) at @xmath83 , represented by a cross in the contour plot . removing the quadrupole
, we obtained similar results , see table 1 for the lower limits on cell sizes .
we obtain the constraint @xmath84 at 95% confidence .
we can not place an upper limit on the cell size : all large cells are equally probable .
the strong constraint from our analysis comes from the predicted power spectrum of the @xmath8 universe ; see figure 1 . according to this plot
, a reduction in the cell size to values below the horizon scale should suppress the quadrupole and low multipole anisotropies , while the suppression is negligible if the cell is very large , at least , larger than the horizon .
it is possible to notice these properties in figures 3 and 4 : both favor large cell sizes .
the observed presence of the quadrupole and other low order anisotropies automatically constrains our cell to be very large . in other words , even before making the @xmath22 fit , we expect to obtain very large cells .
we remind the reader that our analysis is for @xmath1 .
we made this assumption because the results of fitting the @xmath8 model seem to be relatively insensitive to changes in @xmath35 and the bumps " , not the overall slope , are responsible for the poor fit between the model and the data .
in other words , our results are independent of any particular inflationary model . from the cobe / dmr data , we obtain the best @xmath22 fit for a toroidal universe with @xmath85 , which corresponds to a cell size of @xmath86mpc . a cell size below 72% of the size of the horizon ( @xmath87 ) is incompatible with the cobe measurements at 95% confidence , and a cell size below the size of the horizon ( @xmath88 )
is ruled out at 68% confidence .
since the @xmath8 topology is interesting if the cell size is considerably smaller than the horizon , this model loses most of its appeal .
we would like to thank jon aymon , douglas scott , joseph silk , daniel stevens and max tegmark for many useful comments and help with the manuscript .
ac acknowledges sct - pr / cnpq conselho nacional de desenvolvimento cientfico e tecnolgico for her financial support under process no.201330/93 - 8(nv ) . this work was supported in part by the director , office of energy research , office of high energy and nuclear physics , division of high energy physics of the u.s .
department of energy under contract no.de-ac03-76sf00098 .
bond , j.r . & efsthatiou , g. 1987 , mon . not .
, 226:655 .
+ broadhurst , t.j .
et al . 1990 ,
nature , 343:726 .
+ bunn , e. & sugiyama , n. 1994 , preprint ( astro - ph/9407069 ) .
+ ellis , g.f.r .
1971 , gen .
rel . and grav .
+ ellis , g.f.r . &
schreiber , g. 1986 , phys .
a , 115(3):97 .
+ fairall , a.p .
1985 , mon . not .
r. astr . soc . of south .
africa , 44(11):114 .
+ fang , l.z .
& liu , y.l . 1988 ,
a , 3(13):1221 .
+ fang , l.z .
& mo , h. 1987 , mod . phys .
a , 2(4):229 .
+ fang , l.z . &
sato , h. 1985 , gen .
rel . and grav .
, 17(11):1117 .
+ grski , k.m .
1994 , ap .
, 430:l85 .
+ jing , y.p . &
fang , l.z . , 1994 , phy .
, 73(14):1882 .
+ kolb , e.w .
& turner , m.s .
1990 , the early universe , addison - wesley .
+ lineweaver , c. et al .
1994 , ap .
j. , 436:452 .
+ peebles , p.j.e .
1982 , ap .
+ smoot , g.f .
1992 , ap .
+ sokolov , i.y .
1993 , jetp lett .
, 57(10):617 .
+ starobinsky , a.a .
1993 , jetp lett .
, 57(10):622 .
+ stevens , d. et al .
1993 , phys .
, 71(1):20 .
+ wright , e.l .
1994a , ap .
j. , 420:1 .
+ wright , e.l .
1994b , ap .
j. , 436:433 .
+ zeldovich , ya b. 1973 , comm .
space sci .
, 5(6):169 .
+ zeldovich , ya b. and starobinsky , a.a .
1984 , sov .
, 10(3):135 .
+ figure 1 : expected power spectrum for the @xmath8 universe model with @xmath1 for different cell sizes with @xmath11 from 0.1 to 3.0 .
+ figure 2 : power spectrum of the 2-year combined 53 + 90 ghz cobe / dmr data with bias removed and the error bars given by the diagonal terms of the covariance matrix @xmath67 .
+ figure 3 : the likelihood function @xmath78 for the @xmath8 universe model with @xmath1 .
+ figure 4 : the probability that the @xmath8 model is consistent with the data is plotted as a function of the ratio @xmath11 and the normalization @xmath79 ( bottom ) .
confidence limits of 68% , 95% and 99.7% are shown in the contour plot ( top ) .
we found the highest consistency probability ( minimum @xmath22 ) at @xmath85 , represented by a cross in the contour plot . | the cosmic microwave background ( cmb ) is a unique probe of cosmological parameters and conditions .
there is a connection between anisotropy in the cmb and the topology of the universe . adopting a universe with the topology of a 3-torus , or a universe where only harmonics of the fundamental mode are allowed , and using 2-years of cobe / dmr data , we obtain constraints on the topology of the universe .
previous work constrained the topology using the slope information and the correlation function of the cmb .
we obtain more accurate results by using all multipole moments , avoiding approximations by computing their full covariance matrix .
we obtain the best fit for a cubic toroidal universe of scale @xmath0mpc for @xmath1 .
the data set a lower limit on the cell size of @xmath2mpc at 95% confidence and @xmath3mpc at 68% confidence .
these results show that the most probable cell size would be around 1.2 times larger than the horizon scale , implying that the 3-torus topology is no longer an interesting cosmological model .
# 1@xmath4 # 1to 0pt#1 |
let @xmath0 and @xmath1 be two positive @xmath2tuples with@xmath3for each @xmath4 and some constants @xmath5 the following counterparts of the cauchy - bunyakowsy - schwarz inequality are valid : 1 .
plya - szeg s inequality @xcite@xmath6 2 . shisha - mond s inequality @xcite@xmath7 ^{2 } ; \label{1.3}\ ] ] 3 .
ozeki s inequality @xcite@xmath8 4 .
diaz - metcalf s inequality @xcite@xmath9 if @xmath10 is a positive sequence , then the following weighted inequalities also hold : 1 . cassel s inequality @xcite .
if the positive real sequences @xmath11 and @xmath12 satisfy the condition @xmath13then@xmath14 2 .
greub - reinboldt s inequality @xcite@xmath15provided @xmath0 and @xmath1 satisfy the condition @xmath16 3 .
generalised diaz - metcalf inequality @xcite , see also 4b .
if @xmath17 $ ] and @xmath18 @xmath19 and @xmath20 holds , then one has the inequality@xmath21 4 . klamkin - mclenaghan s inequality @xcite . if @xmath22 satisfy ( @xmath23)@xmath24 then@xmath25 for other results providing counterpart inequalities , see the recent monograph on line @xcite .
in this paper we point out a new counterpart of schwarz s inequality in real or complex inner product spaces .
particular cases for isotonic linear functionals , integrals and sequences are also provided .
the following reverse of schwarz s inequality in inner product spaces holds .
[ t2.1]let @xmath26 @xmath27 and @xmath28 if@xmath29or , equivalently,@xmath30holds , then one has the inequality@xmath31the constant @xmath32 is sharp in ( [ 2.2 ] ) .
the equivalence between @xmath33 and @xmath34 can be easily proved , see for example @xcite .
let us define@xmath35\]]and @xmath36then@xmath37 -\left\vert \left\langle x , y\right\rangle \right\vert ^{2}-\left\vert y\right\vert ^{4}\func{re}\left ( a\overline{a}\right)\]]and@xmath38 -\left\vert x\right\vert ^{2}\left\vert y\right\vert ^{2}-\left\vert y\right\vert ^{4}\func{re}\left ( a\overline{a}\right ) , \]]giving@xmath39for any @xmath40 and @xmath41 , which is an interesting equality in itself as well .
if ( [ 2.1 ] ) holds , then @xmath42 and thus@xmath43 .
\label{2.4}\]]if we use the elementary inequality for @xmath44 @xmath45@xmath46 \leq \frac{1}{4}\left\vert u+v\right\vert ^{2 } , \label{2.5}\]]then we have for @xmath47that@xmath48 ^{2}\leq \frac{1}{4}\left\vert a - a\right\vert ^{2}\left\vert y\right\vert ^{4}. \label{2.6}\]]making use of the inequalities @xmath49 and @xmath50 we deduce @xmath51 .
now , assume that ( [ 2.2 ] ) holds with a constant @xmath52 i.e.,@xmath53where @xmath54 satisfy ( [ 2.1 ] ) .
consider @xmath55 @xmath56 @xmath57 and @xmath58 @xmath59 with @xmath60 define@xmath61then@xmath62and thus the condition ( [ 2.1 ] ) is fulfilled . from (
[ 2.61 ] ) we deduce@xmath63and since@xmath64and@xmath65then by ( [ 2.7 ] ) we obtain@xmath66giving @xmath67 and the theorem is completely proved .
let @xmath68 be an algebra of real functions defined on @xmath69 and @xmath70 a subclass of @xmath68 satisfying the conditions : 1 .
@xmath71 implies @xmath72 2 .
@xmath73 @xmath74 implies @xmath75 a functional @xmath76 defined on @xmath77 is an _ isotonic linear functional _ on @xmath77 provided that 1 .
@xmath78 for all @xmath79 and @xmath80 2 .
@xmath81 that is , @xmath82 for all @xmath83 implies @xmath84 the functional @xmath76 is _ normalised _ on @xmath85 provided that @xmath86 i.e. , @xmath87 for all @xmath83 implies @xmath88 usual examples of isotonic linear functionals are integrals , sums , etc . now , suppose that @xmath89 @xmath90 is given and satisfies the properties that @xmath91 @xmath92 @xmath93 for all @xmath94 for a given isotonic linear functional @xmath95 with @xmath96 define the mapping @xmath97
by@xmath98 this functional satisfies the following properties : 1 .
@xmath99 for all @xmath100 2 .
@xmath101 for all @xmath102 and @xmath103 ; 3 .
@xmath104 for all @xmath105 the following reverse of schwarz s inequality for positive linear functionals holds .
[ p3.1]let @xmath106 be such that @xmath91 @xmath107 @xmath108 if @xmath109 are real numbers such that@xmath110then for any isotonic linear functional @xmath95 with @xmath111 we have the inequality@xmath112 ^{2}\leq \frac{1}{4}\left ( m - m\right ) ^{2}a^{2}\left ( hg^{2}\right ) .
\label{3.3}\]]the constant @xmath32 in ( [ 3.3 ] ) is sharp .
we observe that @xmath113 \geq 0.\]]applying theorem [ t2.1 ] for @xmath114 we get@xmath115which is clearly equivalent to ( [ 3.3 ] ) .
the following corollary holds .
[ c3.2]let @xmath116 such that @xmath117 @xmath118 if @xmath109 are real numbers such that ( [ 3.21 ] ) holds , then@xmath119the constant @xmath32 is sharp in ( [ 3.4 ] ) .
[ r3.3]the condition ( [ 3.21 ] ) may be replaced with the weaker assumption@xmath120
let @xmath121 be a measure space consisting of a set @xmath122 a @xmath123algebra @xmath124 of subsets of @xmath125 and a countably additive and positive measure on @xmath124 with values in @xmath126 denote by @xmath127 the hilbert space of all @xmath128-valued functions @xmath129 defined on @xmath125 that are @xmath130integrable on @xmath122 i.e. , @xmath131 where @xmath132 is a measurable function on @xmath133 the following proposition contains a counterpart of the weighted cauchy - buniakowsky - schwarz s integral inequality .
[ p4.1]let @xmath26 @xmath27 and @xmath134 if @xmath135 \rho \left ( s\right ) d\mu \left ( s\right ) \geq 0 \label{4.1}\]]or , equivalently,@xmath136then one has the inequality@xmath137 follows by theorem [ t2.1 ] applied for the inner product @xmath138,@xmath139 [ r4.2]a sufficient condition for ( [ 4.1 ] ) to hold is @xmath140 \geq 0\text { \ for \ } \mu -\text{a.e . }
s\in \omega . \label{4.3}\ ] ] in the particular case @xmath141 we have the following counterpart of the cauchy - buniakowsky - schwarz inequality .
[ c4.3]let @xmath41 @xmath27 and @xmath134 if@xmath135 d\mu \left ( s\right ) \geq 0 , \label{4.4}\]]or , equivalently@xmath142then one has the inequality@xmath143 [ r4.4]if @xmath144 then a sufficient condition for either ( [ 4.1 ] ) or ( [ 4.4 ] ) to hold is @xmath145where , in this case , @xmath146 with @xmath147
for a given sequence @xmath148 of nonnegative real numbers , consider the hilbert space @xmath149 @xmath150 where @xmath151 | a new counterpart of schwarz s inequality in inner product spaces and applications for isotonic functionals , integrals and sequences are provided . |
the different types of core - collapse supernovae ( ccsne ) are produced by the gravitational collapse of the nuclei of massive stars ( @xmath6 ) at the end of their lives .
some of these ccsne are unusually faint and may arise from different types of explosion mechanism and , even , other types of objects misclassified as sne .
the massive eruptions of luminous blue variables ( lbvs ) have similarities with some observed faint , low - velocity , interacting sne @xcite . + ccsne are a heterogeneous class of object demonstrating a wide range of explosion energies and expansion velocities .
@xcite consider a separate branch of subluminous , low velocity type iip sne , such as 1997d , 1999br and 2005cs .
sne with low explosion energies are considered to be produced by either massive stars forming black holes , which quench the sn explosion , or low mass progenitors giving rise to onemg cores and electron capture sne @xcite .
studies of the progenitors of these subluminous sne favour the latter scenario , however , which have been shown to arise from low mass progenitors @xcite .
low velocities are also observed in the spectra of type iin sne , for which various causes have been proposed .
observations of sne 1995 g and 1998s show that in some cases the low - velocity spectral features arise from the interaction between the ejecta and a dense circumstellar medium ( csm ) .
spectroscopy of some type iin sne reveals 2 - 3 velocity components in their strong balmer lines .
@xcite observed , for example , broad , intermediate and narrow components to @xmath5 emission in spectra of 1995 g .
type iin sne span a large range in brightness from sn 1997cy , at maximum @xmath7 @xcite , to sn 2000ch , @xmath8 @xcite .
the type iin sne 1997cy and 1999e are possibly associated with gamma ray bursts ( grbs 970514 and 980910 ; @xcite ) .
the combination of extremely low velocities and faintness has led to the suggestion that some type ii sne are misclassified lbvs undergoing outbursts similar to @xmath9 car @xcite .
+ @xcite present a review of lbv characteristics .
a handful of lbvs have been identified in the galaxy and a few extragalactic lbvs , undergoing outbursts , have also been discovered ( e.g. @xcite ) .
the lbv phase is an important stage in the evolution of the most massive stars ( @xmath10 ; @xcite ) with irregular episodes of mass loss , including extreme eruption events .
the ejected mass goes on to form a csm , which may be observed as an lbv nebula ( such as for @xmath11 car ) .
these massive stars will eventually produce a sn which then interacts with this csm @xcite .
such massive stars are also expected to be progenitors of gamma ray bursts when they are in the wolf - rayet phase of their evolution @xcite .
+ @xcite identified sn 1961v in ngc 1058 as being peculiarly faint and having similar characteristics to @xmath11 car ; this was the prototype for his type v class of sne .
a number of other sne have been noted for their peculiar faintness and low velocities and have been suggested to be lbvs : sn 1954j @xcite , 1978k @xcite , 1997bs @xcite , 1999bw @xcite , 2000ch @xcite , 2001ac @xcite and 2002 kg @xcite . some of these objects have been referred to as `` sn impostors '' @xcite . in the case of sn 1961v
it is not completely clear if in fact this object was the outburst of an lbv . the faint , prolonged and erratic light curve and the low velocities @xcite are consistent with an lbv eruption , but @xcite discuss features of sn 1961v , such as the non - thermal radio spectrum , which are more in keeping with a fading sn remnant .
sn 1986j was suggested by @xcite to be the eruption of an @xmath11 car - like variable , although late time observations showed it to be similar to proper sne @xcite .
the decrease in brightness and the disappearance of the identified progenitor star of sn 1997bs @xcite led @xcite to suggest that sn 1997bs might be a sub - luminous type iin sn .
+ studies of ccsne have been enhanced by the identification of progenitors in pre - explosion imaging . in the case of the lbvs
pre - discovery imaging permits a study of the object at its minimum level of activity .
some have been shown to be coincident with previously identified variable stars , such as sn 1954j = ngc 2403-v12 @xcite and sn 2002 kg = ngc 2403-v37 @xcite .
it is not currently clear , however , where the boundary between outbursting lbvs and intrinsically faint type iin sn falls .
the study of two objects designated sne 2002 kg and 2003gm , and their precursors , is presented here .
details of these sne and their host galaxies are presented in table [ tab : snlbv : galtab ] .
+ [ cols="<,^,^,^,^,^,^ " , ] the pre - discovery broad - band photometry of both objects is affected by large uncertainties .
in addition , as shown in section [ sec : snlbv : results ] ( and by @xcite ) sn2002kg / v37 was an unusual emission line object already in pre - discovery phase .
thus , the transformation of colours and magnitudes into luminosity and effective temperature using the colour - effective temperature relationship for normal supergiants to discuss the physical properties of the objects in the hrd does not seem to be useful .
instead , we compare the location of sn2002kg / v37 and sn2003gm in the colour - magnitude plane with those of similar objects . we will also use the photometry of the surrounding stars to estimate the age of the stellar population .
this simpler and very direct approach will provide important constraints on the nature of the objects . the lbv nature of sn 2002kg / v37 has been previously demonstrated by @xcite . the results presented here ,
however , show that post - discovery data alone can be used to determine the lbv nature of such objects .
the spectroscopic study of sn 2002 kg provides a comparison for future lbvs misidentified as sne , which might not be fortunate enough to have been identified as lbvs prior to discovery of the `` sn . ''
+ the photometry of stars within @xmath12 of sn 2002kg / v37 was used to estimate an age ( see fig .
[ fig : snlbv:2002kgcmd ] ) , by comparing their positions on a colour - magnitude diagram with theoretical isochrones ( from the database of non - rotating stellar evolution models of the geneva group and associated isochrones @xcite ) .
a bimodal age distribution is observed composed of young ( @xmath13 ) and old ( @xmath14 ) populations .
stars from these two populations are evenly distributed together over the 12.5 square arcsecond region around sn 2002kg / v37 .
the age of the young stellar population is consistent with the lifetimes of massive stars with @xmath15 , with the lowest age permitted by the uncertainties including the 2.5myr lifetime of @xmath16 stars .
the uncertainty in the age of the younger stellar population arises from the proximity of the low age isochrones to each other .
the age of the older population implies a maximum mass of @xmath17 for the component stars .
this implies that ngc 2403 has undergone a recent epoch of star formation , at the same location as previous generations of stars
. a worthwhile study would be to use the acquired spectra , or expanded broad band photometry , of the nearby clusters to estimate an age , in a similar manner to that of @xcite , to see if these are similar to the age of the young population of `` field '' stars .
if one considers sn 2002kg / v37 to be either an sn or an lbv then the age requirements for both of these phenomena require the precursor to be a member of the younger stellar population .
+ the brightness and colour of the sn 2002kg / v37 and its precursor are shown on the colour magnitude diagrams of fig .
[ fig : snlbv:2002kghrd ] .
the positions of a number of other lbvs , at outburst and during quiescence , are also plotted for comparison on the left - panel of fig .
[ fig : snlbv:2002kghrd ] .
the uncertainty of the pre- and post - discovery colours does not permit an accurate discussion of the relative change in colours before and after discovery .
the location of the precursor is in the region of the colour - magnitude diagram where lbvs in quiescence , such as p cygni , are found .
the evidence for strong @xmath5 emission in pre - discovery imaging is indicative that the precursor of sn 2002kg / v37 was not a standard blue supergiant .
the change in absolute brightness from the pre - discovery to the post - discovery state is small , of the same order as variability previously observed for ngc2403-v37 .
this may be compared with the change of absolute brightness observed for the progenitors of ccsne such as the peculiarly faint sn 1999br , where @xmath18 .
the change in brightness observed for sn 2002kg / v37 between the pre- and post - discovery states is small compared to that of lbvs where , following the scheme of @xcite , such a magnitude difference constitutes a `` large excursion '' rather than an `` eruption . ''
+ sn 2002 kg follows the general trend of s dor variables which are blue during quiescence and proceed to the red upon outburst , as the increase in radius causes a drop in temperature as the outburst occurs at a constant luminosity .
a key test would be to check if the pre - discovery state lies on the s dor strip and the post - discovery state lies on the constant temperature strip for outbursting lbvs on the hr diagram , between which lbvs move as they change from their quiescent state to outburst @xcite . a number of well studied lbvs are observed to lie in either one of these strips or vary between them @xcite .
the placement of lbvs on the hr diagram requires an accurate effective temperature , which can only be determined with a model atmosphere analysis .
+ the photometric observations , reported on fig .
[ fig : snlbv:2002kglc ] , show that sn 2002kg / v37 has maintained its brightness over a period of at least 700 days .
type iip sne may , as a consequence of the retention of the progenitor s massive h envelope , maintain their brightness over a period of @xmath19 days ( e.g. 1999em : @xcite ; @xcite ) , followed by a steep decline following the decay lifetimes of radioactive elements such as @xmath20 .
the light curve of sn 2002 kg is also at odds with the shallow linear decay observed for low velocity , interacting sne such as 1988z @xcite and 1995 g @xcite , which for interacting sne is due to the conversion of kinetic energy to radiation as a consequence of the interaction of the ejecta with the csm .
the prolonged period of brightness is , however , similar to the observed light curve of the peculiar sn 1961v , which plateaued for @xmath21 months @xcite .
this is also similar to the light curves reported for a number of lbvs such as @xmath9 car @xcite , ngc2403-v12(1954j ) @xcite and ngc 2343-v1 @xcite .
sn 1997bs showed , however , a significant drop in the brightness over a similar time frame .
+ the spectroscopic observations of sn 2002kg / v37 showed minor evolution over the interval of 0.97yrs .
this is similar to the long term behaviour of the lbv ngc2363-v1 @xcite but not with normal type ii sne @xcite .
the reduction in the velocity widths of the balmer lines is much smaller than for the type iin sn 1995 g @xcite , inconsistent with a coasting phase .
the spectrum of sn 2002kg / v37 is very similar to that of ngc2363-v1 and the spectrum of sn 1997bs @xcite , such as the relative line strengths and line widths .
the spectra of sn 2002kg / v37 contain many features , albeit narrower , that are found in the spectrum of sn 1995 g .
the strong feii lines in the spectrum are consistent with those observed in lbvs such as ngc2363-v1 and are sometimes assumed to be a classification criterion for identifying lbv outbursts .
there are , however , a large number of feii lines are also observed in the spectra of sn 1995 g @xcite and no feii lines are seen in the `` fake sn '' sn 2000ch @xcite .
sn 2002kg / v37 does show , however , strong lines , either side of @xmath5 , which are not observed in the type iin sn 1995 g @xcite .
these strong lines are indicative of raised nitrogen abundance in an `` lbv nebula , '' formed from ejection of n rich material surrounding the lbv @xcite .
the presence of these lines is important due to the conspicuous absence of other nebular lines , such as those listed in table [ tab : snlbv : hiitab ] , implying a raised n abundance .
the strengths of the lines , relative to @xmath5 , reported here are smaller than those observed in the lbv nebulae studied by @xcite .
this is because sn 2002kg / v37 is not - spatially resolved from the nebula and so , the observations reported here , contain flux from the star in addition to the nebula .
other `` nebular '' lines observed by @xcite in the lbv nebulae of r127 , s119 and r143 show the other `` nebular '' lines are @xmath22 the strength of the .
the strengths of the lines might be used as a classification criterion for lbvs , which have been originally identified as sne .
this requires the fwhm of @xmath5 to be @xmath23@xmath24 , to resolve the line at @xmath25@xmath4 , and to have good signal - to - noise .
this fwhm limit is less than half that observed for the broad component of @xmath5 in early spectra of sn 1995 g @xcite .
the photometric and spectroscopic observations of sn 2002kg / v37 show that this object was and is an lbv , and present template observations with which future lbv / sne candidates can be compared .
the diverse range of phenomena observed for lbvs @xcite means that the more data available for definitively classified objects , the better future classification will be .
having confirmed the lbv nature of sn2002kg / v37 it seems important to obtain accurate information about effective temperature , luminosity and chemical composition .
this can be accomplished through a quantitative spectral analysis of our emission line spectra using spherical extended non - lte line blanketed model atmospheres including stellar winds .
such work for emission line stars in galaxies beyond the local group has been carried out recently by @xcite in ngc 300 and by @xcite for the lbv ngc2363-v1
. this will be the next step of our work .
the acs / hrc images were used to provide photometry of stars within 2@xmath26 of sn 2003gm ( see fig . [
fig : snlbv:2003gmcmd ] ) .
this photometry was plotted on a colour magnitude diagram , with half - solar metallicity isochrones ( appropriately shifted for the reddening , extinction and distance to ngc 5334 = 20.5mpc ) to estimate an age for these stars .
an approximate age of @xmath27 was estimated from this colour - magnitude diagram , which is similar to that determined for sn 1997bs in m66 ( see fig .
[ fig : snlbv:1997bscmd ] ) .
this corresponds to the lifetime of stars with initial masses of @xmath28 . this is lower than the @xmath29 initial mass threshold for stars to undergo an lbv phase in the standard `` conti '' scenario .
the position of sn 2003gm and its precursor on the colour magnitude diagram are compared with stellar evolution tracks on fig .
[ fig : snlbv:2003gmhrd ] .
the precursor appears consistent with a yellow supergiant , with @xmath30 .
the error bars for the position of the precursor also enclose the points of termination of tracks for stars with @xmath31 .
the stellar evolution tracks of the geneva group predict that the sn progenitors of stars with @xmath32 should be red supergiants , with stars in the @xmath33 range being yellower .
in the case of an lbv one might not expect that this colour is representative of the star itself , rather it would be representative of the nebula surrounding the lbv in a similar manner to the homonculus shrouding @xmath11 car . similarly in the case of sn progenitors circumstellar dust will cause the observed colours to deviate from the actual colours of the progenitor , affecting the inferred spectral type of the progenitor .
@xcite reported the discovery of a yellow supergiant progenitor for a , seemingly , normal type iip sn 2004et .
a yellow supergiant was observed as the progenitor of sn 1993j @xcite , although this was due to the heavy stripping of the h envelope by a binary companion .
the brightness of the precursor is similar to that of the precursor of sn 1997bs ( @xmath34 , @xcite ) , although there was no colour information for the latter .
+ the light curve for sn 2003gm ( see fig . [
fig : snlbv:2003gmlc ] ) decays over a similar period to which sn 2002kg / v37 was observed to remain at approximately constant brightness .
caution is required with the interpretation of the light curve of sn 2003gm , which may have varied between observations .
@xcite showed , however , the erratic light curve of an apparent lbv outburst , sn 2000ch , had a sharp decline after maximum light .
@xcite also showed the light curve of sn 1961v was erratic .
another lbv candidate , 1997bs , showed a smooth behaviour post - discovery @xcite .
this suggests that the shape of the light curve immediately post - maximum is not suitable for identifying lbv outbursts originally classified as sne .
+ the absolute unfiltered magnitude at discovery @xmath35 is similar to the brightnesses for the peculiarly faint sn 1999br @xcite and the proposed sn impostors sn 1961v @xcite and sn 1997bs @xcite , but @xmath36mag brighter than sn 2000ch at maximum brightness @xcite . at maximum brightness
this is still @xmath37mags brighter than ngc2363-v1 @xcite , and corresponds to an increase in brightness from the precursor of @xmath38 .
the late time light curve of sn 2003gm is compared with the late time light curves of the `` sn impostor '' sn 1997bs , the peculiarly faint sn 1999br and type iin sn 1995
[ fig : snlbv:2003gmcomplc ] .
+ the two early time spectra of sn 2003gm appear very similar to spectra of sn 2000ch @xcite , composed of a featureless continuum with large balmer emission lines .
the host galaxy of sn 2003gm is twice the distance of the host of sn 2000ch and hence the spectra are noisier than those published of sn 2000ch . similarly to sn 2000ch lines are not observed , as might be expected in lbv outbursts , but this may be a consequence of the poor signal to noise rather than the absence of .
the widths of the balmer lines are observed to narrow between the first two epochs of spectroscopy .
the fading of sn 2003gm , i.e. the way in which sn features have faded to reveal the spectrum of an hii region , is similar to the way in which sn 1997bs faded @xcite .
the manner in which both sn 2003gm and sn 1997bs faded contradicts the suggested rule of @xcite : that the precursors of lbvs should remain visible after their outburst .
the spectrum of sn 1997bs , presented by @xcite , shows similar balmer features to sn 2003gm , as well as showing significantly detected feii lines , which would probably not have been detected in the sn 2003gm spectra due to the levels of noise .
+ the last spectrum shows features consistent with an ionised hii region .
this ionized region could be the host for the progenitor of an sn , ionized prior to explosion , or could be unrelated .
the metallicity determination for sn 2003gm shows no abundance anomalies . a comparison of line strength ratios from the last spectrum of the site of sn 2003gm with the locus of normal hii regions on the @xmath396583\aa / h\alpha)}$ ] vs. @xmath405007\aa / h\beta)}$ ] diagram , as presented by @xcite as their fig .
1 , shows no unusual line strengths .
the line strength ratios are inconsistent with those observed from the forbidden lines of seyfert and liner galaxies , environments which host large shocks that might be expected when sn ejecta collides with the csm .
the ratio of the 6583@xmath4line strength to that of @xmath5 suggests the ionised region is not an lbv nebula . in an lbv nebula one expects to see the skewed n abundance due to the mass loss of n rich material as the lbv was undergoing outburst and the elevated levels of mass loss expected for lbvs during minimum .
+ the combination of low luminosity and low velocities suggest that if sn 2003gm is indeed a sn then it is a very under - energetic one .
@xcite predict that progenitors with @xmath41 may form black holes and are , hence , quenched .
sn 2003gm is similar to sn 1997bs , but an authoritative argument for the lbv nature of the latter has not been put forward .
post - discovery photometric and spectroscopic observations of two faint , low velocity outbursts have been presented here .
+ the post - discovery observations of sn 2002kg / v37 are consistent with the conclusion of @xcite , who identified sn 2002 kg as being the previously identified lbv ngc2403-v37 .
the photometric behaviour of sn 2002 kg has been shown to be inconsistent with the behaviour of normal interacting type ii sn , maintaining its approximate brightness over a period of just under 2 years .
the spectroscopy of sn 2002kg / v37 shows it to be similar to other identified lbvs , such as ngc2363-v1 .
strong lines were observed in the spectrum , in common with other lbvs , and this feature is suggested for the classification of misidentified lbvs .
the results of this work show that the lbv nature of a sn candidate may be determined over a long period .
if the lines can be used to identify lbvs , then this work also shows that these features are present in very early spectra and can lead to a classification without the need for observations over an extended period . +
the nature of sn 2003gm is unclear , although it is photometrically and spectroscopically similar to sn 1997bs .
the photometric behaviour of both of these objects , from the limited observations , could fall within the `` general envelope '' of the behaviour expected for lbvs .
the lbv characteristic features observed in sn 2002kg / v37 were not observed in sn 2003gm .
future observations of both sn 1997bs and sn 2003gm will show if either of these objects are recurrent variable stars .
high resolution deep imaging of the sites of sne 1997bs and 2003gm will show whether the precursor object has disappeared , and long term observations , similar to those of ngc2403-v37 @xcite , will show if these objects are recurrent .
+ progenitor observations have permitted the determination of the relative increase in brightness of the precursor to the objects discovered as sn 2002kg / v37 and sn 2003gm . in the case of the former the increase in brightness
is small . in the case of sn
2003gm the increase in brightness , while modest , is significantly brighter than most lbvs .
+ velocity and brightness criteria can not be used as a definitive classification technique .
sn 2002kg / v37 was of sufficient faintness as to make it extremely unlikely to be a sn .
sn 2003gm was , however of similar brightness to sn 1997bs and only slightly fainter than sn 1999br at the time of discovery .
strict brightness and velocities limits on the classification scheme would preclude the possible identification of potential sub - luminous type iin sne , rather identifying them as lbvs .
+ spectroscopic classification of these objects with , for example , the lines is a promising way of distinguishing between the two types of events in the future .
an archive of observations of these objects , for example made available through the suspect website ] , would permit careful comparison of the properties of such objects discovered and help provide definitive classification criteria in the future .
jr acknowledges financial support , in the form of a postdoctoral fellowship , from the university of texas at austin .
sjs thanks the euryi scheme for a fellowship and financial support .
some of data presented here was made publicly available through the isaac newton groups wide field camera survey programme .
some of the data presented here were obtained with the william herschel telescope .
the isaac newton telescope and william herschel telescopes are operated on the island of la palma by the isaac newton group in the spanish observatorio del roque de los muchachos of the instituto de astrofisica de canarias .
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 .
this work uses observations made with the nasa / esa hubble space telescope , obtained from the data archive at the space telescope science institute .
stsci is operated by the association of universitiesfor research in astronomy , inc . under nasa contract
nas 5 - 26555 .
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 made use of the hyperleda databases ( @xmath42 ) .
some sne spectra were retrieved from the suspect online supernova spectrum archive .
jrm thanks p.a .
crowther , j. fabbri , m. irwin , t. kim , a.d .
mackey , l.j .
smith and j.c .
wheeler for useful discussion .
l. , ford h. c. , huchra j. , kennicutt r. c. , mould j. r. , sakai s. , freedman w. l. , stetson p. b. , madore b. f. , gibson b. k. , graham j. a. , hughes s. m. , illingworth g. d. , kelson d. d. , macri l. , sebo k. , silbermann n. a. , 2000 , , 128 , 431 r. m. , vrba f. j. , henden a. a. , canzian b. , luginbuhl c. b. , filippenko a. v. , chornock r. , li w. , coil a. l. , schmidt g. d. , smith p. s. , starrfield s. , klose s. , tich ' a j. , tich ' y m. , gorosabel j. , hudec r. , simon v. , 2004 , , 116 , 326 | photometric and spectroscopic observations of the faint supernovae ( sne ) 2002 kg and 2003gm , and their precursors , in ngc 2403 and ngc 5334 respectively , are presented .
the properties of these sne are discussed in the context of previously proposed scenarios for faint sne : low mass progenitors producing under - energetic sne ; sne with ejecta constrained by a circumstellar medium ; and outbursts of massive luminous blue variables ( lbvs ) .
the last scenario has been referred to as `` type v sne '' , `` sn impostors '' or `` fake sne . '' + the faint sn 2002 kg reached a maximum brightness of @xmath0 , much fainter than normal type ii sne .
the precursor of sn 2002 kg is confirmed to be , as shown in previous work , the lbv ngc2403-v37 .
late time photometry of sn 2002 kg shows it to be only 0.6 magnitudes fainter at 500 days than at the epoch of discovery .
two spectra of sn 2002 kg , with an approximately 1 year interval between observations , show only minor differences .
strong @xmath1 lines are observed in the spectra of sn 2002 kg , similar to both the lbv ngc2363-v1 and the type iin sn 1995 g .
the spectrum of sn 2002 kg does show strong resolved @xmath2}$ ] at @xmath36549,6583@xmath4 .
the identified progenitor of sn 2003gm is a bright yellow star , consistent with a f5-g2 supergiant , similar to the identified progenitor of sn 2004et .
sn 2003gm , at the epoch of discovery , was of similar brightness to the possible fake sn 1997bs and the type iip sne 1999br and 2005cs .
photometrically sn 2003gm shows the same decrease in brightness , over the same time period as sn 1997bs . the light curve and the spectral properties of sn 2003gm are also consistent with some intrinsically faint and low velocity type ii sn .
the early time spectra of sn 2003gm are dominated by balmer emission lines , which at the observed resolution , appear similar to sn 2000ch .
on the basis of the post - discovery photometric and spectroscopic observations presented here we suggest that sn 2003gm is a similar event to sn 1997bs , although the sn / lbv nature of both of these objects is debated . at 226 days post
- discovery the spectrum of sn 2003gm is strongle contaminated by hii region emission lines , and it can not be confirmed that the precursor star has disappeared .
the presence of strong @xmath2}$ ] lines , near @xmath5 , is suggested as a possible means of identifying objects such as sn 2002kg / ngc2403-v37 as being lbvs - although not as a general classification criterion of all lbvs masquerading as sne .
stars : variables : other stars : individual : ngc2403-v37 supernovae : general supernovae : individual : 2002 kg
supernovae : individual : 2003gm galaxies : individual : ngc 5334 |
the statistical significance associated with the detection of a signal source is most often reported in the form of a @xmath0-value , that is , the probability under the background - only hypothesis of observing a phenomenon as or even more ` signal - like ' than the one observed by the experiment . in many simple situations
, a @xmath0-value can be calculated using asymptotic results such as those given by wilk s theorem @xcite , without the need of generating a large number of pseudo - experiments .
this is not the case however when the procedure for detecting the source involves a search over some range , for example , when one is trying to observe a hypothetic signal from an astrophysical source that can be located at any direction in the sky .
wilk s theorem does not apply in this situation since the signal model contains parameters ( i.e. the signal location ) which are not present under the null hypothesis .
estimation of the @xmath0-value could be then performed by repeated monte carlo simulations of the experiment s outcome under the background - only hypothesis , but this approach could be highly time consuming since for each of those simulations the entire search procedure needs to be applied to the data , and to establish a discovery claim at the @xmath1 level ( @xmath0-value=@xmath2 ) the simulation needs to be repeated at least @xmath3 times .
fortunately , recent advances in the theory of random fields provide analytical tools that can be used to address exactly such problems , in a wide range of experimental settings .
such methods could be highly valuable for experiments searching for signals over large parameter spaces , as the reduction in necessary computation time can be dramatic .
random field theoretic methods were first applied to the statistical hypothesis testing problem in @xcite , for some special case of a one dimensional problem . a practical implementation of this result , aimed at the high - energy physics community ,
was made in @xcite .
similar results for some cases of multi - dimensional problems @xcite@xcite were applied to statistical tests in the context of brain imaging @xcite .
more recently , a generalized result dealing with random fields over arbitrary riemannian manifolds was obtained @xcite , openning the door for a plethora of new possible applications . here
we discuss the implementation of these results in the context of the search for astrophysical sources , taking icecube @xcite as a specific example . in section [ sec1 ]
the general framework of an hypothesis test is briefly presented with connection to random fields . in section [ sec2 ] the main theoretical result
is presented , and an example is treated in detail in section [ sec3 ] .
the signal search procedure can be formulated as a hypothesis testing problem in the following way .
the null ( background - only ) hypothesis @xmath4 , is tested against a signal hypothesis @xmath5 , where @xmath6 represents the signal strength .
suppose that @xmath7 are some nuisance parameters describing other properties of the signal ( such as location ) , which are therefore not present under the null .
additional nuisance parameters , denoted by @xmath8 , may be present under both hypotheses .
denote by @xmath9 the likelihood function .
one may then construct the profile likelihood ratio test statistic @xcite @xmath10 and reject the null hypothesis if the test statistic is larger then some critical value .
note that when the signal strength is set to zero the likelihood by definition does not depend on @xmath7 , and the test statistic ( [ eq : q ] ) can therefore be written as @xmath11 where @xmath12 is the profile likelihood ratio with the signal nuisance parameters fixed to the point @xmath7 , and we have explicitely denoted by @xmath13 the @xmath14-dimensional manifold to which the parameters @xmath7 belong . under the conditions of wilks theorem @xcite , for any fixed point @xmath7 , @xmath12 follows a @xmath15 distribution with one degree of freedom when the null hypothesis is true . when viewed as a function over the manifold @xmath13 , @xmath12 is therefore a @xmath15 _ random field _ , namely a set of random variables that are continuously mapped to the manifold @xmath13 . to quantify the significance of a given observation in terms of a @xmath0-value , one is required to calculate the probability of the maximum of the field to be above some level ,
that is , the excursion probability of the field : @xmath16.\ ] ] estimation of excursion probabilities has been extensively studied in the framework of random fields . despite the seemingly difficult nature of the problem
, some surprisingly simple closed - form expressions have been derived under general conditions , which allow to estimate the excursion probability ( [ eq : pval ] ) when the level @xmath17 is large .
such ` high ' excursions are of course the main subject of interest , since one is interested in estimating the @xmath0-value for apparently significant ( signal - like ) fluctuations .
we shall briefly describe the main theoretical results in the following section . for a comprehensive and precise definitions ,
the reader is referred to ref .
the excursion set of a field above a level @xmath17 , denoted by @xmath18 , is defined as the set of points @xmath7 for which the value of the field @xmath12 is larger than @xmath17 , @xmath19 and we will denote by @xmath20 the _ euler characteristic _ of the excursion set @xmath18 . for a 2-dimensional field , the euler characteristic can be regarded as the number of disconnected components minus the number of ` holes ' , as is illustrated in fig.[fig : eulerillus ] . a fundamental result of @xcite states that the expectation of the euler characteristic @xmath20 is given by the following expression : @xmath21 = \sum_{d=0}^d \mathscr{n}_d \rho_d(u).\ ] ] the coefficients @xmath22 are related to some geometrical properties of the manifold and the covariance structure of the field . for the purpose of the present analysis however
they can be regarded simply as a set of unknown constants .
the functions @xmath23 are ` universal ' in the sense that they are determined only by the distribution type of the field @xmath12 , and their analytic expressions are known for a large class of ` gaussian related ' fields , such as @xmath15 with arbitrary degrees of freedom .
the zeroth order term of eq .
( [ eq : euler ] ) is a special case for which @xmath24 and @xmath25 are generally given by @xmath26\ ] ] namely , @xmath24 is the euler characteristic of the entire manifold and @xmath25 is the tail probability of the distribution of the field .
( note that when the manifold is reduced to a point , this result becomes trivial ) .
when the level @xmath17 is high enough , excursions above @xmath17 become rare and the excursion set becomes a few disconnected hyper - ellipses . in that case
the euler characteristic @xmath20 simply counts the number of disconnected components that make up @xmath18 .
for even higher levels this number is mostly zero and rarely one , and its expectation therefore converges asymptotically to the excursion probability .
we can thus use it as an approximation to the excursion probability for large enough @xmath17 @xcite @xmath27 \approx \mathbb{p } [ \displaystyle\max_{\theta \in \mathscr{m } } q(\theta ) > u].\ ] ] the practical importance of eq .
( [ eq : euler ] ) now becomes clear , as it allows to estimate the excursion probabilities above high levels .
furthermore , the problem is reduced to finding the constants @xmath28 . since eq .
( [ eq : euler ] ) holds for any level @xmath17 , this could be achieved simply by calculating the average of @xmath20 at some low levels , which can be done using a small set of monte carlo simulations .
we shall now turn to a specific example where this procedure is demonstrated .
the icecube experiment @xcite is a neutrino telescope located at the south pole and aimed at detecting astrophysical neutrino sources .
the detector measures the energy and angular direction of incoming neutrinos , trying to distinguish an astrophysical point - like signal from a large background of atmospheric neutrinos spread across the sky .
the nuisance parameters over which the search is performed are therefore the angular coordinates @xmath29 .
we follow @xcite for the definitions of the signal and background distributions and the likelihood function .
the signal is assumed to be spatially gaussian distributed with a width corresponding to the instrumental resolution of @xmath30 , and the background from atmospheric neutrinos is assumed to be uniform in azimuthal angle .
we use a background simulation sample of 67000 events , representing roughly a year of data , provided to us by the authors of @xcite .
we then calculate a profile likelihood ratio as described in the previous section .
figure [ fig : map ] shows a `` significance map '' of the sky , namely the values of the test statistic @xmath31 as well as the corresponding excursion set above @xmath32 . to reduce computation time we restrict here the search space to the portion of the sky at declination angle 27@xmath33 below the zenith , however all the geometrical features of a full sky search
are maintained .
note that the most significance point has a value of the test statistic above 16 , which would correspond to a significance exceeding 4@xmath34 if this point would have been analyzed alone , that is without the `` look elsewhere '' effect . in practice
, the test statistic @xmath31 is calculated on a grid or points , or ` pixels ' , which are sufficiently smaller than the detector resolution .
the computation of the euler characteristic can then be done in a straightforward way , using euler s formula : @xmath35 where @xmath36 , @xmath37 , and @xmath38 are respectively the numbers of _ vertices _ ( pixels ) , _ edges _ and _ faces _ making up the excursion set .
an edge is a line connecting two adjacent pixels and a face is the square made by connecting the edges of four adjacent pixels .
an illustration is given fig.[fig : illus2 ] .
( although it is most convenient to use a simple square grid , other grid types can be used if necessary , in which case the faces would be of other polygonal shapes ) . ) .
each square represents a pixel . here
, the number of vertices is 18 , the number of edges is 23 and the number of faces is 7 , giving @xmath39 . , width=453 ] once the euler characteristic is calculated , the coefficients of eq .
( [ eq : euler ] ) can be readily estimated . for a @xmath15 random field with one degree of freedom and for two search dimensions , the explicit form of eq .
( [ eq : euler ] ) is given by @xcite : @xmath40 = \mathbb{p}[\chi^2 > u ] + e^{-u/2}(\mathscr{n}_1 + \sqrt{u}\mathscr{n}_2).\ ] ] to estimate the unknown coefficients @xmath41 we use a set of 20 background simulations , and calculate the average euler characteristic of the excursion set corresponding to the levels @xmath42 ( the number of required simulations depends on the desired accuracy level of the approximation . for most practical purposes ,
estimating the @xmath0-value with a relative uncertainty of about 10% should be satisfactory . ) .
this gives the estimates @xmath43=33.5 \pm 2 $ ] and @xmath44=94.6 \pm 1.3 $ ] . by solving for the unknown coefficients we obtain @xmath45 and @xmath46 .
the prediction of eq .
( [ eq : ex1 ] ) is then compared against a set of approx . 200,000 background simulations , where for each one the maximum of @xmath31 is found by scanning the entire grid .
the results are shown in figure [ fig:2 ] . as expected , the approximation becomes better as the @xmath0-value becomes smaller .
the agreement between eq .
( [ eq : ex1 ] ) and the observed @xmath0-value is maintained up to the smallest @xmath0-value that the available statistics allows us to estimate . )
( dashed red ) against the observed @xmath0-value ( solid blue ) from a set of 200,000 background simulations .
the yellow band represents the statistical uncertainty due to the available number of background simulations .
, width=302,height=264 ] a useful property of eq .
( [ eq : euler ] ) that can be illustrated by this example , is the ability to consider only a small ` slice ' of the parameter space from which the expected euler characteristic ( and hence @xmath0-value ) of the entire space can be estimated , if a symmetry is present in the problem .
this can be done using the ` inclusion - exclusion ' property of the euler characteristic : @xmath47 since the neutrino background distribution is assumed to be uniform in azimuthal angle ( @xmath48 ) , we can divide the sky to @xmath49 identical slices of azimuthal angle , as illustrated in figure [ fig : slice ] .
applying ( [ eq : slicing ] ) to this case , the expected euler characteristic is given by @xmath50 = n\times(\mathbb{e}[\phi(slice)]-\mathbb{e}[\phi(edge ) ] ) + \mathbb{e}[\phi(0)]\ ] ] where an ` edge ' is the line common to two adjacent slices , and @xmath51 is the euler characteristic of the point at the origin ( see figure [ fig : slice ] ) . ) . in this example
@xmath52 , @xmath53 and @xmath54.,width=302 ] we can now apply eq .
( [ eq : euler ] ) to both @xmath55 and @xmath56 and estimate the corresponding coefficients as was done before , using only simulations of a single slice of the sky . following this procedure
we obtain for this example with @xmath57 slices from 40 background simulations , @xmath58 and @xmath59 .
using ( [ eq : euler_slice ] ) this leads to the full sky coefficients @xmath60 and @xmath61 , a result which is consistent with the full sky simulation procedure .
this demonstrates that the @xmath0-value can be accurately estimated by only simulating a small portion of the search space .
the euler characteristic formula , a fundamental result from the theory of random fields , provides a practical mean of estimating a @xmath0-value while taking into account the `` look elsewhere effect ''
. this result might be particularly useful for experiments that involve a search for signal over a large parameter space , such as high energy neutrino telescopes .
while the example considered here deals with a search in a 2-dimensional space , the formalism is general and could be in principle applied to any number of search dimensions .
for example , if one is trying to detect a ` burst ' event then time would constitute an additional search dimension .
in such case the method of slicing could be useful as well , as one will not have to simulate the entire operating period of the detector but only a small slice of time ( provided that the background does not vary in time ) .
thus , the computational burden of having to perform a very large number of monte carlo simulations in order to to estimate a @xmath0-value , could be greatly reduced .
we thank jim braun and teresa montaruli for their help in providing us the background simulation data of icecube which was used to perform this analysis .
one of us ( e. g. ) is obliged to the minerva gesellschaft for supporting this work .
davies , _ hypothesis testing when a nuisance parameter is present only under the alternative .
_ , biometrika * 74 * ( 1987 ) , 33 - 43 .
e. gross and o. vitells , _ trial factors for the look elsewhere effect in high energy physics _ , eur . phys .
j. c , * 70 * ( 2010 ) , 525 - 530 . r.j .
adler and a.m. hasofer , _ level crossings for random fields_,ann .
probab . * 4 * , number 1 ( 1976 ) , 1 - 12 .
adler , _ the geometry of random fields _ , new york ( 1981 ) , wiley , isbn : 0471278440 .
worsley , s. marrett , p. neelin , a.c .
vandal , k.j .
friston and a.c .
evans , _ a unified statistical approach for determining significant signals in location and scale space images of cerebral activation _
, human brain mapping * 4 * ( 1996 ) 58 - 73 .
adler and j.e .
taylor , _ random fields and geometry _ , springer monographs in mathematics ( 2007 ) .
isbn : 978 - 0 - 387 - 48112 - 8 .
j. ahrens et al . and the icecube collaboration , astropart . phys .
* 20 * ( 2004 ) , 507 .
j. braun , j. dumma , f. de palmaa , c. finleya , a. karlea and t. montaruli , _
methods for point source analysis in high energy neutrino telescopes _ , astropart .
* 29 * ( 2008 ) 299 - 305 [ arxiv:0801.1604 ] . | in experiments that are aimed at detecting astrophysical sources such as neutrino telescopes , one usually performs a search over a continuous parameter space ( e.g. the angular coordinates of the sky , and possibly time ) , looking for the most significant deviation from the background hypothesis .
such a procedure inherently involves a `` look elsewhere effect '' , namely , the possibility for a signal - like fluctuation to appear anywhere within the search range . correctly estimating the @xmath0-value of a given observation thus requires repeated simulations of the entire search , a procedure that may be prohibitively expansive in terms of cpu resources .
recent results from the theory of random fields provide powerful tools which may be used to alleviate this difficulty , in a wide range of applications .
we review those results and discuss their implementation , with a detailed example applied for neutrino point source analysis in the icecube experiment .
look - elsewhere effect , statistical significance , neutrino telescope , random fields |
the system under study is described by the following second quantization hamiltonian @xmath0 the energy dispersions of the two quantum well conduction subbands are @xmath1 and @xmath2 , being @xmath3 the electron in - plane wavevector and @xmath4 the effective mass ( non - parabolicity is here neglected ) .
the corresponding electron creation fermionic operators are @xmath5 and @xmath6 , where @xmath7 is the electron spin .
@xmath8 is the bare frequency dispersion of a cavity photonic branch as a function of the in - plane wavevector @xmath9 , where @xmath10 is the light speed , @xmath11 is the cavity spacer dielectric constant and @xmath12 is the quantized photon wavevector along the normal direction . @xmath13
is the corresponding photon creation operator , obeying bosonic commutation rules . due to the well - known polarization selection rules of intersubband transitions ,
we omit the photon polarization , which is assumed to be transverse magnetic ( tm ) . for simplicity , we consider only a photonic branch , which is quasi - resonant with the intersubband transition , while other cavity photon modes are supposed to be off - resonance and can be therefore neglected in first approximation .
the interaction between the cavity photon field and the two electronic subbands is quantified by the coupling constant @xmath14 where @xmath15 is the intersubband transition dipole along the quantum well growth direction , @xmath16 is the frequency of the intersubband transition , @xmath17 the vacuum permittivity , @xmath18 is the effective cavity length and @xmath19 is the sample area . for simplicity ,
we have considered a @xmath20-cavity , with @xmath21 being the quantized vector along the growth direction . the geometrical factor @xmath22 originates from the tm - polarization nature of the transition . moreover , in eq .
( [ chi ] ) we have assumed that the active quantum well is located at the antinode of the cavity mode field , providing maximum coupling . note that here we have neglected the anti - resonant terms of the light - matter interaction and therefore we can describe the strong coupling regime for the electrically excited system , but not the ultrastrong coupling limit , as instead done in refs . .
the hamiltonian term @xmath23 is meant to include all the other interactions : ( i ) electron - phonon interaction ; ( ii ) electron - electron interaction ; ( iii ) electron tunneling coupling to the injection and extraction reservoir ; ( iv ) coupling between the cavity photon field and the extracavity field .
it is known that due to the cubic light - matter coupling term in the hamiltonian ( the product of two fermion operators and one boson operator ) it is not possible to write down an exact closed set of equations for the evolution of operators , being the heisenberg equation of motion for each product of @xmath24 operators coupled at least with one product of @xmath25 operators . in other words ,
the equations of motion of the different observables of the system form an infinite hierarchy .
one approximation method that has been used in order to solve this kind of systems is the so - called cluster expansion scheme@xcite .
it is based on a systematic development of expectation values of operator products in terms of correlation functions . in order to obtain a consistent truncation scheme
, a pair of fermionic operators has to be considered of the same order as a single bosonic operator . in this work
, we have truncated the hierarchy at the level of the product of two excitation operators ( i.e. , the product of four fermion operators ) .
the details of the factorization are in appendix [ app ] .
the expectation values entering the present cluster factorization are the electronic and photonic populations , the correlation between the cavity photon field and the intersubband polarization , as well as polarization - polarization correlations .
the electron occupation numbers in the two quantum well conduction subbands are @xmath26 and @xmath27 .
note that , since in the absence of a magnetic field all quantities are spin - independent , we omit the spin - index in the notation of the averaged quantities .
the cavity photon number is @xmath28 .
the correlation between the cavity photon field and the intersubband electronic polarization is represented by the quantity @xmath29 finally , the polarization - polarization correlation function is given by @xmath30 note that in the spontaneous photon emission regime , @xmath31 can not be factorized : in fact , spontaneous emission is incoherent and @xmath32 , @xmath33 , meaning that the cavity field and the intersubband polarization have no definite phase .
loss of coherence due to dephasing processes and photonic losses is phenomenologically quantified by the damping rate @xmath34 .
unlike @xmath31 , @xmath35 can be factorized in products of non - zero lower - order expectation values of operators .
in fact , we have @xmath36 .
the first contribution is an uncorrelated plasma term , while @xmath37 describes the higher - order correlation , which can be destroyed by dephasing processes quantified by the damping rate @xmath38 .
the terms in @xmath23 , namely the phonon scattering , electron - electron interaction , the coupling to the contact reservoirs and the coupling to the external electromagnetic field will be treated in an effective way .
the carrier non - radiative relaxation ( due to phonon - electron and electron - electron scattering ) is modeled in terms of a simple phenomenological relaxation time @xmath39 .
note that the role of coulomb electron - electron interaction on intersubband transitions has been studied , e.g. , in ref . .
in the case of subbands with parallel parabolic dispersion ( e.g. , same effective mass ) , coulomb interaction produces a moderate renormalization of the intersubband transition frequency @xmath40 and of its oscillator strength , which will not be accounted explicitly in the present work .
let @xmath41 and @xmath42 be the self - consistent local equilibrium occupation numbers .
they are given by fermi - dirac distributions : @xmath43 where @xmath44 is the boltzmann thermal factor , and @xmath45 is the quantum well self - consistent fermi level , such that : @xmath46 the two subbands are coupled to two electronic reservoirs , named respectively left and right contacts .
we will call @xmath47 the electronic tunneling rate into the @xmath3-mode of the subband @xmath48 from the reservoir @xmath49 .
analogously @xmath50 is defined as the electronic tunneling rate from the @xmath3-mode of the subband @xmath51 into the reservoir @xmath52 .
the total in - tunneling and out - tunneling rates are @xmath53 and @xmath54 .
the resulting closed system of equations for the one - time expectation values reads : @xmath55 sketch of the energy dispersion of the two quantum well subbands and of the minibands in the left and right contacts in the zero - bias case . here
the system is in thermal equilibrium and the fermi level in the quantum well is the same as in the two contacts .
the doping level in the contacts determines the equilibrium density in the quantum well .
the subband and minibands have an energy dispersion versus the in - plane wavevector , which is a conserved quantity in the planar structure .
this electronic structure is embedded in a planar microcavity , with a cavity photon mode quasi - resonant to the intersubband transition.,width=302 ] same as in fig .
[ nobias ] , but with an applied voltage bias . here
, the left contact acts as an electronic extractor , while the right one is the injector . in the quantum
well , non - equilibrium steady - state populations can be established in the two subbands .
, width=302 ] the wave - vector dependent injection and extraction rates in eq .
( [ full_system ] ) can be in principle of different origin . here
we give the formal expression for elastic tunneling processes conserving the in - plane momentum .
additional processes ( such as assisted tunneling ) can be accounted for by adding their contribution to the expressions for @xmath56 and @xmath57 to be inserted in eq .
( [ full_system ] ) . as electronic contact reservoirs , we will consider semiconductor doped superlattices , as it is generally the case in unipolar quantum cascade devices .
the chemical potential in each contact is labeled @xmath58 with @xmath49 . in each reservoir
, we will consider miniband states with energy @xmath59 . in the elastic tunneling process ,
electron energy and in - plane momentum are conserved .
the tunneling rate from the contact reservoir into the j - th subband is @xmath60 where @xmath61 is the tunneling matrix element and @xmath62 is in general an index over the electronic states of the miniband with in - plane wave vector @xmath3 .
it can be interpreted as the axial electronic wave vector in the case the two leads are just bulk contacts .
@xmath63 is the fermi - dirac occupation number of the electron states in the contact .
analogously the tunneling rate from the j - th subband of the quantum well into the reservoir @xmath52 reads @xmath64 where @xmath65 is the hole occupation number in the contact .
the value of @xmath66 can be quantum engineered , depending on the specific structure . in particular , by changing the thickness of the potential barriers , it is possible to tailor considerably the tunneling matrix element . it is straightforward to see that a simple relationship occurs between @xmath67 and @xmath68 , namely @xmath69 note that here we have assumed that the bare energy dispersion of the electrons in the two subbands is unaffected .
this is valid in the weak light - matter coupling regime or when the injector miniband energy width is broad enough . for large values of the vacuum rabi frequency ,
the spectral function of the electrons in the second subband is non - trivially modified as well as the tunneling process using a narrow - band injector .
this will be proved and discussed in detail in a forthcoming paper@xcite .
in this work , we will focus on the steady - state solutions for the quantities @xmath70 , @xmath71 , @xmath72 , @xmath31 and @xmath73 . hence
, we can set the time derivatives equal to zero , transforming the differential system ( [ full_system ] ) into an algebraic one . in the steady - state regime ,
the electronic current ( number of electrons per unit time ) through the structure is given by the expression : @xmath74 the total rate of photons emitted out of the microcavity reads @xmath75 where @xmath76 is the escape time of a photon out of the microcavity .
the _ quantum efficiency _
@xmath77 is defined as the ratio between the photonic current out of the cavity and electronic current , i.e. , @xmath78 .
in the steady - state regime , the momentum - dependent spontaneous photon emission spectra are given by the expression : @xmath79 in order to determine @xmath80 , we need to solve the following heisenberg equations of motion @xmath81 note that here we have omitted the coupling of the electronic injector and extractor reservoirs to the quantity @xmath82
. this coupling would involve correlations between the quantum well electronic field and the contact electronic fields . since in this paper
we are dealing with incoherent electron transport , we will neglect such correlations with the contact reservoirs , which are also extremely tricky to tackle . truncating the hierarchy at the level of two excitations ( details in appendix [ algebraic ] ) and taking the unilateral fourier transform ( @xmath83 ) we obtain : @xmath84 where @xmath85 , @xmath86 and @xmath87 represents half the difference between the total number of electrons in the fundamental subband and the number in the second one , namely : @xmath88 note that the total density of electrons is @xmath89 , where the @xmath90 factor accounts for the two - fold spin degeneracy of the electron states in the conduction subbands . @xmath91 and @xmath92 are phenomenological damping rates for @xmath93 and @xmath94 respectively .
the analytical solutions are @xmath95 @xmath96 the electroluminescence spectrum is simply @xmath97 from the analytical result for @xmath98 , we see immediately that emission spectrum is resonant at the two polariton frequencies @xmath99 satisfying the equation @xmath100 the quantity @xmath101 is just the vacuum rabi frequency of the present system . at resonance ( i.e. , @xmath102 ) , the necessary condition for the appearance of a strong coupling polaritonic splitting is @xmath103 , meaning that the total density of electrons in the fundamental subband must be larger enough than the total density in the second . for a vacuum rabi frequency much larger than @xmath104 and @xmath105 , the minimum polariton splitting is given by twice the vacuum rabi frequency .
note that here the electroluminescence spectral shape does not depend explicitly on the spectral properties of the injector and extractor reservoirs .
the spectrum in eq .
( [ el ] ) has the same shape as the absorption ( in presence of the same carrier densities ) .
the dependence on the transport is only implicit , being given by the steady - state carrier and photon populations .
in contrast , in the exact solution of the the simplified model of ref . , it is shown that the electroluminescence spectra is the absorption spectrum times the spectral distribution of excitations in the electronic reservoir , which then acts as an electronic filter@xcite . a fermionic approach based on an exact diagonalization method
@xcite indeed shows that the spectral properties of the electronic contact modifies significantly the spectral shape of the electroluminescence in the case of narrow band injectors .
hence , the spectrum predicted by eq .
( [ el ] ) is valid only for broad band injectors .
this is not really surprising because , in order to calculate the tunneling rates , we have considered bare electronic states in the quantum well and have only considered incoherent population injection and extraction processes .
current density versus applied voltage for different values of the intersubband transition energy : @xmath106mev , ( dashed - dotted line ) , @xmath107 mev ( dashed line ) and @xmath108mev ( solid line ) .
other parameters can be found in the text.,width=302 ] photonic current density versus electronic current for different values of the intersubband transition energy : @xmath109mev , ( dashed - dotted line ) , @xmath110mev ( dashed line ) and @xmath108mev ( solid line ) .
same parameters and range of applied voltages as in fig .
[ iv - curve].,width=302 ] electron occupation number in the fundamental conduction subband as a function of kinetic energy and applied voltage .
inset : the integrated density of electrons in the fundamental subband versus voltage . @xmath111mev and other parameters as in fig .
[ iv - curve ] . for @xmath112 ,
the density of electrons in the first subband is @xmath113 .
, width=302 ] same as in fig .
[ n1 ] , but for the second subband .
inset : the integrated density of electrons in the second subband versus voltage . for @xmath112 ,
the density of electrons in the second subband is @xmath114 .
, width=302 ] contour plot of the photon occupation ( log scale ) versus the applied voltage and the energy of the bare cavity photon mode .
, width=302 ] [ numerical ] here , we apply our theory using realistic parameters for a microcavity - embedded quantum cascade electroluminescent source . in order to simplify the algebra ,
we have systematically neglected the photon wavevector whenever added to an electronic wavevector . given
the huge difference in the typical wavectors of photons and electrons , this simplification is safe . applying this approximation ,
we can obtain a closed set of algebraic equation where the variables are the populations in the two subbands and in the cavity photonic branch , as shown in appendix [ algebraic ] .
this system has been solved numerically using a standard newton method .
we achieve numerical convergence in a relatively fast computation time except in the limit of vanishing bias , when the injector and extractor are strongly misaligned with the two subbands . physically in this case the steady - state situation is reached in times very long compared to the dynamics of the quantum well system , the photon population is extremely small and correspondingly the numerical method fails to converge .
anyway this is not a real limitation , because we are interested in the behavior of the system in presence of a finite voltage bias , producing a significant current flow and photonic output . in figs .
[ nobias ] and [ bias ] , we show a sketch of the energy profile of the injector and extractor with respect to the quantum well subbands respectively without and with an applied bias .
specifically , in the numerical calculations we have used the following electronic out - tunneling rates : @xmath115 where @xmath116 , @xmath117ps , @xmath118 and @xmath119 are the energy offsets of the left and right minibands .
the in - tunneling rates are determined by applying the relation in eq .
( [ rapporto ] ) .
in all the simulations , we have taken @xmath120 and @xmath121 .
note that these are just phenomenological injection rates . for the amplitude @xmath122 , we have considered values which are consistent with what realistically obtainable in semiconductor intersubband devices .
importantly , in real structures @xmath122 can be considerably quantum engineered by changing the barrier thickness and/or the miniband structure of the injection superlattices .
this is why we have not considered a very specific injector configuration and taken the simplified expression in eq .
( [ injector ] ) with realistic parameters .
when a voltage bias is applied , the two reservoirs are shifted symmetrically , as shown in fig.[bias ] . in all the simulations , except when otherwise stated we used the realistic damping parameters @xmath123 , @xmath124 , while the temperature is @xmath125k . in the simulations we have also considered @xmath39 to be independent from @xmath3 and such that @xmath126 , except when otherwise stated .
note that here we have considered only an active quantum well .
for quantum cascade structures with several active quantum wells repeated in a periodic way , the dynamics is similar and the present treatment can be generalized without major difficulties . in the simulations ,
the intersubband transition energy @xmath127 is , except where otherwise stated , equal to @xmath108mev and the coupling constant @xmath128 is such that the vacuum rabi frequency is @xmath129 for an electron density of @xmath130@xmath131 ( all in the fundamental subband ) .
when @xmath132 is changed , the coupling constant is adjusted in order to keep the ratio between the vacuum rabi frequency and transition frequency constant . the effective mass @xmath133 has been taken to be one tenth of the bare electronic mass . in the numerical calculations ,
the cavity spacer dielectric constant is @xmath134 . for each simulation
, the resonance in - plane wavector @xmath135 , given by the condition @xmath136 , corresponds to an internal cavity photon propagation angle @xmath137 equal to 70 degrees , where @xmath138 . in fig .
[ iv - curve ] , we show the current density versus applied voltage ( between the injector and extractor ) for different values of @xmath132 .
the current - voltage profile is characteristic of an unipolar quantum cascade light emitting diode .
the current grows superlinearly in the voltage region where the injector fermi level approaches the second subband .
the current is bigger for smaller @xmath132 because , keeping the injection rate @xmath122 constant ( but all the internal rates of the system proportional to @xmath132 ) , the injection and extraction processes become the dominant processes .
note that an increase of the nonradiative relaxation rate @xmath139 produces a nearly proportional increase of the electronic current ( not shown ) .
the rates of emitted photons per unit area ( integrated all over the in - plane wavevectors ) are shown in fig .
[ photons ] as a function of the flowing current , showing an approximately linear behavior .
[ n1 ] and [ n2 ] show contour plots of the electron occupation numbers of the first and second subband respectively as a function of the applied voltage and of the kinetic energy .
the insets in fig .
[ n1 ] and [ n2 ] show respectively the integrated density of electrons in the first and second subband .
it is apparent that with increasing voltage the population in the first subband decreases , while the population in the second subband increases .
when the injector fermi level becomes aligned with the second subband , as expected , the carrier occupation numbers in the two subbands are considerably out of equilibrium .
the decrease of the first subband carrier occupation numbers is beneficial for the radiative efficiency of the spontaneous emission , because the influence of pauli blocking is reduced .
moreover , in the considered conditions , the density of electrons in the first subband is still considerably larger than in the second subband , thus producing a large vacuum rabi coupling and efficient emission rate .
[ photon_k ] contains a contour plot of the cavity photon occupation number versus the bare photon energy , showing that the maximum of emission is obtained when the bare photon energy is resonant with the intersubband transition , as expected and as observed experimentally @xcite . with the considered parameters , the density of electrons in the first subband is high enough to be in the strong coupling regime , as depicted in fig .
[ spectrum ] , where the anticrossing of two polariton branches is clearly present .
the minimum polariton splitting , given by the expression @xmath140 is reported in fig .
[ splitting ] as a function of the applied bias . with increasing voltage
, the population difference @xmath141 diminishes .
this results in a decrease of the vacuum rabi frequency and consequently of the polariton splitting .
this high - excitation feature has been already observed in experiments@xcite and can not be described within a bosonic approach@xcite , which can be applied only to the low excitation density case@xcite .
it is interesting to analyze the quantum efficiency @xmath77 , defined as the ratio between the photonic emission rate and the electronic current , namely @xmath142 in fig .
[ efficiencygamma ] , we plot the quantum efficiency @xmath77 at @xmath112 versus the vacuum rabi frequency @xmath143 at the same voltage ( log - log scale ) . in the simulations ,
the vacuum rabi frequency has been varied by changing the coupling constant @xmath144 . in a realistic quantum engineered device ,
@xmath144 can be tailored in different ways .
for example , by growing the active quantum wells in a spatial region where the cavity mode field is very small , it is possible to quench dramatically the value of @xmath144 .
moreover , by using different shape of quantum wells , it is also possible to tailor the transition dipole @xmath15 .
[ efficiencygamma ] shows that in the weak coupling regime ( small values of @xmath145 ) the efficiency grows like @xmath146 . in the strong coupling regime ,
the efficiency becomes impressive and then tends to saturate .
it is apparent that the radiative efficiency smoothly increases passing from the weak to the strong coupling regime .
this crossover occurs because the radiative efficiency depends on the spectrally integrated emission and it is therefore insensitive to the sudden appearence of the polariton doublet in the strong coupling emission spectra .
this results are in qualitative agreement with the analytical solutions of the simplified model in ref .
, where only the bright intersubband states are considered and where the electronic reservoir is modeled with a bath of harmonic oscillators . as shown in fig .
[ efficiencytau ] , the nonradiative population relaxation rate @xmath139 has the most significant effect . in the considered regime of parameters ,
the efficiency is proportional to @xmath147 .
it is interesting to compare our results for this microcavity system with the standard free space case . in the free - space case ,
the photon current , obtained by applying the fermi golden rule , is given by the formula @xmath148 .
as it is well known , the free - space radiative efficiency dramatically decreases with the intersubband emission wavelength due to the @xmath149 dependence of the spontaneous emission rate ( @xmath150 , so the spontaneous emission rate scales effectively as @xmath151 ) . in the mid - infrared , by using the same parameters , for a transition of @xmath108 mev , the quantum efficiency is of the order of @xmath152 .
hence , it is clear from our results that a strong coupling light - emitting diode based on a planar microcavity system can provide a dramatic enhancement with respect to the free space case ( even three orders of magnitude for the larger vacuum rabi frequency case ) .
contour plot of the electroluminescence ( arb .
units ) as a function of the bare cavity photon energy @xmath153 and of the emission frequency @xmath154 for an applied voltage @xmath155 .
the anticrossing of the two intersubband polariton branches is apparent in electroluminescence spectra.,width=340 ] minimum polariton splitting as a function of the applied voltage . @xmath111mev and
other parameters can be found in the text.,width=340 ] quantum efficiency versus the corresponding vacuum rabi frequency at the voltage @xmath156 .
the three lines are obtained with different values of the coherence damping coefficients : @xmath157 ( solid line ) , @xmath158 ( dashed line ) and @xmath159 ( dashed - dotted line ) .
@xmath111mev.,width=351 ] quantum efficiency versus the corresponding vacuum rabi frequency at the voltage @xmath156 .
the three lines are obtained with different values of the non - radiative relaxation rate : @xmath160 ( dashed line ) , @xmath161 ( solid line ) and @xmath162 ( dashed - dotted line ) .
, width=351 ]
in conclusion , we have presented a quantum theoretical study of the quantum well intersubband electroluminescence from a semiconductor microcavity in the incoherent transport regime , i.e. when the coupling to the electronic contacts concerns only the electron populations .
the problem has been tackled starting from the fermionic electron hamiltonian for the two - subbands and by using a cluster factorization method to truncate the infinite hierarchy of dynamical equations for the relevant expectation values of operator products . at the present level of approximation
, we have been able to describe the incoherent electron transport through the dynamics of the electronic subband populations and the electroluminescence through the dynamics of the cavity photon population , which is coupled to the correlations between the electromagnetic field and the intersubband polarization .
we have discussed the limits of applicability of the present approach , which neglects the impact of the vacuum rabi coupling on the quantum well carrier spectral function and any correlation between the quantum well and the contact reservoirs .
the analogies and differences with the exact predictions of the simplified model in ref . have been critically and extensively discussed .
we have shown the appearance of cavity polariton resonances in the emission spectra , when a large density of electrons occupies the fundamental subband .
we have described how the vacuum rabi splitting decreases with increasing voltage and described the photonic output in the different transport conditions .
our results show that even in presence of non - radiative relaxation and pauli blocking , the quantum efficiency of the microcavity intersubband electroluminescence can be considerably enhanced by increasing the vacuum rabi frequency .
a more refined treatment@xcite based on a fermionic exact diagonalization method shows that under certain conditions the strong vacuum rabi coupling regime affect considerably not only the dynamics of the intersubband polarization ( hence the absorption spectrum ) , but also the quantum well electron spectral properties and consequently the tunneling transport using narrow - band injectors . as future perspective
, this could be exploited to further improve the quantum efficiency of microcavity intersubband emitters and for the eventual realization of intersubband polariton lasers .
we are pleased to thank i. carusotto , r. colombelli , l. sapienza , c. sirtori , a. vasanelli for discussions .
as stated in the main body of the paper we used a cluster expansion and truncation scheme to obtain a closed set of equations . here we briefly review the principles of this method following @xcite and apply it to the actual case .
if we consider each bosonic operator or each pair of fermionic operators as an excitation operator and we write the expectation value of an @xmath24 excitation operator as @xmath163 , then the heisenberg equation of motion takes the form : @xmath164 where the n - excitation expectation value is coupled to higher order quantities via the functional v. an n - excitation truncation scheme is obtained if we factorize all the expectation values of more than @xmath24 excitation in all the possible ways and considering the sign exchange for the fermionic operators in order to obtain a factorized quantity that respects the commutation and anticommutation properties of the original quantity .
we are interested in incoherent emission only , so the only nonzero one excitation operators we consider are @xmath165 and @xmath166 .
we factorized the @xmath167 excitations operators in the following way : @xmath168 for the two - time quantities in the calculation of luminescence spectrum , we proceed analogously and obtain : @xmath169
in the steady - state regime , neglecting the photonic wavevector into sums over electronic wavevectors , the system of equations ( [ full_system ] ) reduces to the following system of algebraic equations . discretizing the electronic and photonic wavevectors on a grid of respectively @xmath175 and @xmath176 points
, we have a system of @xmath177 equations that can be numerically solved , e.g. , with a newton algorithm . , edited by h. c. liu and f. capasso , semiconductors and semimetals vol .
62 ( academic press , san diego , 2000 ) .
d. dini , r. kohler , a. tredicucci , g. biasiol , and l. sorba , phys .
lett . * 90 * , 116401 ( 2003 ) .
a. a. anappara , a. tredicucci , g. biasiol , l. sorba , appl .
phys . lett . *
87 * , 051105 ( 2005 ) .
a. a. anappara , a. tredicucci , f. beltram , g. biasiol , l. sorba , appl .
phys . lett . * 89 * , 171109 ( 2006 ) . c. ciuti , g. bastard , i. carusotto , phys .
b * 72 * , 115303 ( 2005 ) . c. ciuti , i. carusotto , phys . rev .
a * 74 * , 033811 ( 2006 ) . s. de liberato , c. ciuti , i. carusotto , phys .
lett . * 98 * , 103602 ( 2007 ) .
j. faist , f. capasso , d.l .
sivco , c. sirtori , a. l. hutchinson , a.y .
cho , science * 264 * , 5553 ( 1994 ) .
r. khler , a. tredicucci , f. beltram , h.e .
beere , e.h .
linfield , a.g .
davies , d.a .
ritchie , r.c .
iotti , f. rossi , nature * 417 * , 156 ( 2002 )
. r. colombelli , k. srinivasan , m. troccoli , o. painter , c. f. gmachl , d. m. tennant , a. m. sergent , d. l. sivco , a. y. cho , f. capasso , science * 302 * , 1374 ( 2003 ) .
r. colombelli , c. ciuti , y. chassagneux , c. sirtori , semicond .
technol . * 20 * , 985 ( 2005 ) .
l. sapienza , a. vasanelli , c. ciuti , c. manquest , c. sirtori , r. colombelli , and u. gennser , appl .
. lett . * 90 * , 201101 ( 2007 ) .
l. sapienza , a. vasanelli , r. colombelli , c. ciuti , y.chassagneux , c. manquest , u. gennser , c. sirtori , phys .
lett . * 100 * , 136806 ( 2008 ) .
s. de liberato and c. ciuti , arxiv:0802.4091 .
j. fricke , ann .
phys . * 252 * , 479 ( 1996 ) .
j. fricke , v. meden , c.wohler , k. schonhammer , ann .
phys . * 253 * , 177 ( 1997 ) .
m. kira , w. hoyer , t. stroucken , s. w. koch , phys .
lett . * 87 * , 176401 ( 2001 ) .
l. sapienza , ( ph . d. thesis , university of paris 7 , 2007 ) .
d. e. nikonov , a. imamoglu , l. v. butov , and h. schmidt , phys .
79 * , 4633 ( 1997 ) .
m. f. pereira , jr .
b * 75 * , 195301 , ( 2007 ) . | we present a quantum theoretical analysis of the electroluminescence from an intersubband transition of a quantum well structure embedded in a planar microcavity . by using a cluster factorization method ,
we have derived a closed set of dynamical equations for the quantum well carrier and cavity photon occupation numbers , the correlation between the cavity field and the intersubband polarization , as well as polarization - polarization contributions . in order to model the electrical excitation ,
we have considered electron population tunneling from an injector and into an extractor contact .
the tunneling rates have been obtained by considering the bare electronic states in the quantum well and the limit of validity of this approximation ( broad - band injection ) are discussed in detail .
we apply the present quantum model to provide a comprehensive description of the electronic transport and optical properties of an intersubband microcavity light emitting diode , accounting for non - radiative carrier relaxation and pauli blocking .
we study the enhancement of the electroluminescence quantum efficiency passing from the weak to the strong polariton coupling regime and compare it with the free - space case . in the last two decades , the fundamental research on the physics of intersubband transitions in semiconductor quantum wells has enjoyed a considerable success and also led to novel applications in quantum optoelectronics@xcite . recently ,
reflectivity experiments @xcite have demonstrated that by embedding a doped quantum well structure in a planar microcavity , it is possible to achieve the strong coupling regime between an intersubband transition and a cavity photon mode , provided that a dense enough two - dimensional electron gas populates the fundamental quantum well subband . the interaction between a bright intersubband excitation and a cavity photon is quantified by the so - called vacuum rabi frequency .
the strong coupling regime occurs when the vacuum rabi frequency exceeds the electronic and photonic losses .
in such a regime , the normal modes of the system are cavity polaritons , half - photon half - intersubband excitations . in this kind of system
, it is even possible to reach an unconventional ultra - strong coupling regime , i.e. a vacuum rabi frequency comparable to the intersubband transition frequency@xcite .
the interplay between judiciously quantum engineered intersubband transitions and vertical electron transport is the essence of the so - called quantum cascade electroluminescent devices and lasers , which are unipolar optoelectronic sources emitting in the mid and far infrared portion of the electromagnetic spectrum @xcite .
a new kind of microcavity - embedded quantum cascade devices in the strong coupling regime was proposed in ref . .
the first experimental demonstrations of a microcavity quantum cascade photovoltaic detector@xcite and of an electroluminescent device in the strong coupling regime have been recently reported@xcite .
this promising research topic is in its very infancy and many interesting theoretical questions need to be addressed . in ref .
, intersubband polariton electroluminescence has been analytically treated within a simplified hamiltonian model based on the following assumptions : ( i ) only the bright intersubband excitations have been taken into account , while dark excitations have been neglected ; ( ii ) only the low excitation regime has been considered , in which the bright intersubband excitations have been approximated as bosons ; ( iii ) the electronic coupling to the intersubband polarization field has been modeled through a phenomenological reservoir of bosonic excitations . in this work , we will attempt to treat the same problem starting directly from the fermionic hamiltonian for the quantum well carriers . this approach can give us useful insight to understand the physics obtained by relaxing the assumptions used in ref . and to grasp which intrinsic factors ultimately determine the quantum efficiency of these strong coupling emitters . on one hand , the large values of the vacuum rabi frequency could induce a very fast and efficient emission of photons . on the other hand , the large density of dark intersubband excitations created by the injection current and the pauli blocking in the densely populated fundamental subband could suppress such enhancement .
we would like to point out that from a theoretical point of view , a description of the considered system in terms of the fermionic carrier operators makes the system hilbert space much larger than within a bosonic model .
in this paper , we have followed an approach based on a truncation of the infinite hierarchy of dynamical equations for the operator expectation values , allowing us to describe many relevant aspects of the intersubband microcavity electroluminescence .
however , some of the non - perturbative features obtained analytically within a bosonic model@xcite can not be accounted for within the present treatment .
different fermionic approaches based on exact diagonalization methods @xcite are eventually necessary for further refinements . in this paper
, we present a quantum model of the spontaneous photon emission from an electrically - excited intersubband transition of a quantum well structure embedded in a planar microcavity mode .
here , we will consider the case of an incoherent electron transport , where the quantum well electron populations in the two subbands have a tunneling coupling to an electronic injector and to an extractor .
the tunneling rates have been obtained by considering the bare electron states inside the quantum well .
the domain of validity of this approximation will be discussed in detail .
the present theoretical model is applied to describe the incoherent electron transport and electroluminescence of an intersubband microcavity light emitting diode in the strong coupling regime .
the paper is structured as follows . in sec .
[ ham ] , we describe the system and introduce the second quantization hamiltonian describing electrons in the two conduction subbands and photons in the fundamental microcavity mode . in sec .
[ onetime ] , we present a closed set of dynamical equations for the one - time expectation values of operator products , describing photon and carrier populations as well as intersubband polarization - polarization and polarization - field correlations .
these equations have been obtained through a cluster expansion , whose details are reported in appendix [ app ] . in sec .
[ steady ] , we discuss the steady - state regime obtained under constant electrical excitation .
the corresponding set of algebraic equations for the steady - state expectation values are reported in appendix [ algebraic ] . in sec .
[ spectra ] , the electroluminescence spectra are analytically calculated as a function of the populations and the appearance of intersubband cavity polaritonic resonances in the emission spectra is shown .
numerical applications of the theory are presented in sec .
[ numerical ] , using a specific configuration for the injection and extraction electronic reservoirs .
the results predict the current - voltage characteristics , emission spectra and quantum efficiency using different ( controllable ) parameters for the considered microcavity system .
the results are critically discussed with respect to the approximations of the model . finally , conclusions and future perspectives are drawn in sec . [ conclusions ] |
this paper fully describes the sequences given by the non - autonomous second order lyness difference equations @xmath4 where @xmath2 is a @xmath5-periodic sequence taking positive values , @xmath6 and the initial conditions @xmath3 are as well positive .
this question is proposed in ( * ? ? ?
5.43 ) . recall that non - autonomous recurrences appear for instance as population models with a variable structure affected by some seasonality @xcite , where @xmath5 is the number of seasons .
some dynamical issues of similar type of equations have been studied in several recent papers @xcite .
recall that when @xmath7 that is @xmath8 for all @xmath9 , then ( [ eq ] ) is the famous lyness recurrence which is well understood , see for instance @xcite .
the cases @xmath10 have been already studied and some partial results are established . for both cases
it is known that the solutions are persistent near a given @xmath5-periodic solution , which is stable .
this is proved by using some known invariants , see @xcite .
recall that in our context it is said that a solution @xmath0 is persistent if there exist two real positive constants @xmath11 and @xmath12 which depend on the initial conditions , such that for all @xmath13 we prove : [ main ] let @xmath0 be any sequence defined by ( [ eq ] ) and @xmath14 .
then it is persistent .
furthermore , either 1 .
the sequence @xmath0 is periodic , with period a multiple of @xmath5 ; or 2 .
the sequence @xmath0 densely fills one or two ( resp .
one , two or three ) disjoint intervals of @xmath15 when @xmath2 is 2-periodic ( resp .
3-periodic ) .
moreover it is possible by algebraic tools to distinguish which is the situation .
our approach to describe the sequences @xmath0 is based on the study of the natural dynamical system associated to ( [ eq ] ) and on the results of @xcite .
the main tool that allows to distinguish the number of intervals for the adherence of the sequences @xmath0 is the computation of several resultants , see section [ provaa ] .
it is worth to comment that theorem [ main ] is an extension of what happens in the classical case @xmath16 .
there , the same result holds but in statement ( b ) only appears one interval .
our second main result will prove that there are other more significative differences between the case @xmath16 and the cases @xmath17 these differences are related with the lack of monotonicity of certain rotation number functions associated to the dynamical systems given by the lyness recurrences , see theorem [ teonomonotonia ] .
the behaviors of these rotation number functions are important for the understanding of the recurrences , because they give the possible periods for them , see @xcite .
on the other hand in @xcite it is proved that , at least for some values of @xmath18 the behaviour of @xmath0 for the case @xmath19 is totally different . in particular unbounded positive solutions
appear . in the forthcoming paper @xcite
we explore in more detail the differences between the cases @xmath20 and @xmath21 this paper is organized as follows : section [ ds ] presents the difference equations that we are studying as discrete dynamical systems and we state our main results on them , see theorems [ rotacions ] and [ teonomonotonia ]
. section [ pt2 ] is devoted to the proof of theorem [ rotacions ] . by using it , in section [ provaa ] , we prove theorem [ main ] and we give some examples of how to apply it to determine the number of closed intervals of the adherence of @xmath0 . in section [ someproperties ]
we demonstrate theorem [ teonomonotonia ] and we also present some examples where we study in more detail the rotation number function of the dynamical systems associated to .
in this section we reduce the study of the sequence @xmath0 to the study of some discrete dynamical systems and we state our main results on them .
first we introduce some notations . when @xmath22 set @xmath23 and when @xmath24 set @xmath25 where @xmath26 and @xmath27 and @xmath28 we also consider the maps @xmath29 , with @xmath30 as @xmath31 defined on the open invariant set @xmath32 consider for instance @xmath33 the sequence given by ( [ eq ] ) , @xmath34 can be seen as @xmath35 hence the behavior of ( [ seqq ] ) can be obtained from the study of the dynamical system defined in @xmath36 by the map : @xmath37 similarly , for @xmath38 we can consider the map : @xmath39 notice that both maps have an only fixed point in @xmath40 which depends on @xmath41 ( and @xmath11 ) , that for short we denote by @xmath42 it is easy to interpret the invariants for ( [ eq ] ) and @xmath6 given in @xcite , in terms of first integrals of the above maps , see also lemma [ elemental ] .
we have that @xmath43 is a first integral for @xmath44 and @xmath45 is a first integral for @xmath46 .
the topology of the level sets of these integrals in @xmath36 as well as the dynamics of the maps restricted to them is described by the following result , that will be proved in section [ pt2 ] .
[ rotacions ] * the level sets of @xmath47 ( resp .
@xmath48 ) in @xmath49 are diffeomorphic to circles surrounding @xmath50 , which is the unique fixed point of @xmath44 ( resp .
@xmath46 ) . * the action of @xmath44 ( resp .
@xmath46 ) on each level set of @xmath47 ( resp .
@xmath48 ) contained in @xmath49 is conjugated to a rotation of the circle .
once a result like the above one is established the study of the possible periods of the sequences @xmath0 given by ( [ eq ] ) is quite standard .
it suffices , first to get the rotation interval , which is the open interval formed by all the rotation numbers given by the above theorem , varying the level sets of the first integrals .
afterwards , it suffices to find which are the denominators of all the irreducible rational numbers that belong to the corresponding interval , see @xcite .
the study of the rotation number of these kind of rational maps is not an easy task , see again @xcite . in particular , in @xcite was proved that the rotation number function parameterized by the energy levels of the lyness map @xmath51 is always monotonous , solving a conjecture of zeeman given in @xcite , see also @xcite .
as far as we know , in this paper we give the first simple example for which this rotation number function is neither constant nor monotonous .
we prove : [ teonomonotonia ] there are positive values of @xmath52 and @xmath53 , such that the rotation number function @xmath54 of @xmath44 associated to the closed ovals of @xmath55 has a local maximum . hence , apart from the known behaviors for the autonomous lyness maps , that is global periodicity or monotonicity of the rotation number function , which trivially holds for @xmath56 taking for instance @xmath57 or @xmath58 respectively , there appear more complicated behaviors for the rotation number function .
our proof of this result relies on the study of lower and upper bounds for the rotation number of @xmath44 on a given oval of a level set of @xmath47 given for some @xmath59 and @xmath60 for @xmath61 .
this can be done because the map on this oval is conjugated to a rotation and it is possible to use an algebraic manipulator to follow and to order a finite number iterates on it , which are also given by points with rational coordinates .
so , only exact arithmetic is used .
a similar study could be done for @xmath46 .
_ proof of ( i ) of theorem [ rotacions ] . _
the orbits of @xmath44 and @xmath46 lie on the level sets @xmath62 and @xmath63 respectively .
these level sets can be seen as the algebraic curves given by @xmath64 and @xmath65 respectively .
taking homogeneous coordinates on the projective plane @xmath66 both curves @xmath67 and @xmath68 have the form @xmath69 in order to find the branches of them tending to infinity , we examine the directions of approach to infinity ( @xmath70 ) in the local charts determined by @xmath71 and @xmath72 respectively . in the local chart given by @xmath71 ,
the curve @xmath73 writes as @xmath74 and it meets the straight line at infinity @xmath70 when @xmath75 since for both curves @xmath67 and @xmath68 the coefficients @xmath76 and @xmath77 are positive , the only intersection point that could give points in @xmath36 is @xmath78 the algebraic curve @xmath73 arrives to @xmath79 tangentially to the line @xmath80 since for both curves , @xmath67 and @xmath81 the coefficients @xmath76 and @xmath82 are also positive , we have that the branches of the level sets tending to infinity are not included in @xmath83 an analogous study can be made in the chart given by @xmath84 obtaining the same conclusions .
moreover , it can be easily checked that in the affine plane both curves @xmath67 and @xmath68 do not intersect the part of the axes @xmath85 and @xmath86 which is in the boundary of @xmath36 . in summary , there are no branches of the curves @xmath67 and @xmath68 tending to infinity or crossing the axes @xmath85 and @xmath86 in @xmath87 , and therefore the connected components of @xmath88 for @xmath89 are bounded .
notice that this result in particular already implies the persistence of the sequences given by .
consider @xmath90 .
we claim the following facts : a. in @xmath36 , the set of fixed points of @xmath44 and the set of singular points of @xmath91 coincide and they only contain the point @xmath92 .
b. the function @xmath93 has a local minimum at @xmath50 .
we remark that item ( b ) is already known .
we present a new simple proof for the sake of completeness . from the above claims and the fact that the connected components of the level sets of @xmath47 in @xmath36 are bounded it follows that the level sets of @xmath47 in @xmath94 are diffeomorphic to circles .
let us prove the above claims .
the fixed points of @xmath44 are given by @xmath95 and so @xmath96 hence in @xmath97 we have that @xmath98 and the above system is equivalent to @xmath99 it is not difficult to check that the last system of equations is precisely the one that gives the critical points of the curves @xmath100 moreover , from the first equation it is necessary that @xmath101 and hence @xmath102 since @xmath103 has only one real root in @xmath104 the uniqueness of the critical point holds .
let us prove that this critical point corresponds with a local minimum of @xmath105 we will check the usual sufficient conditions given by the hessian of @xmath47 at @xmath50 .
firstly , @xmath106 secondly , the determinant of the hessian matrix at the points @xmath107 is @xmath108 where @xmath109 a tedious computation shows that @xmath110 with @xmath111 observe that if @xmath112 is the positive root of @xmath103 , then @xmath113 .
taking into account that @xmath114 implies that @xmath115 we have that @xmath116 so , @xmath117 since @xmath118 @xmath119 and , on this interval , there is only one critical point of @xmath120 which is simple , we get that @xmath121 and so @xmath122 hence @xmath123 is a local minimum of @xmath124 , as we wanted to prove .
the same kind of arguments work to end the proof for the case @xmath24 but the computations are extremely more tedious .
we only make some comments .
the fixed points of @xmath46 in @xmath87 are given by : @xmath125 it can be proved again that they coincide with the singular points of @xmath48 in @xmath83 this fact follows from the computation of several suitable resultants between @xmath126 @xmath127 and @xmath128 the uniqueness of the fixed point @xmath50 in @xmath36 can be shown as follows : since @xmath129 implies that @xmath130 , we only need to search solutions of @xmath131 in @xmath132 . with the new variable @xmath133 @xmath134 since @xmath135 ; @xmath136 ; and the descarte s rule , we know that there is only one positive solution , as we wanted to see .
finally it can be proved that @xmath50 is a non - degenerated local minimum of @xmath48 .
these computations are complicated , and they have been performed in a very smart way in @xcite , so we skip them and we refer the reader to this last reference . ' '' '' in @xcite it is proved a result that characterizes the dynamics of integrable diffeomorphisms having a _ lie symmetry _ , that is a vector field @xmath137 such that @xmath138 .
next theorem states it , particularized to the case we are interested .
[ teor ] let @xmath139 be an open set and let @xmath140 be a diffeomorphism such that : 1 .
it has a smooth regular first integral @xmath141 having its level sets @xmath142 as simple closed curves .
2 . there exists a smooth function @xmath143 such that for any @xmath144 @xmath145 then the map @xmath146 restricted to each @xmath147 is conjugated to a rotation with rotation number @xmath148 , where @xmath149 is the period of @xmath147 as a periodic orbit of the planar differential equation @xmath150 and @xmath151 is the time needed by the flow of this equation for going from any @xmath152 to @xmath153 next lemma is one of the key points for finding a lie symmetry for families of periodic maps , like the @xmath154 and @xmath155periodic lyness maps .
[ mus ] let @xmath156 be a family of diffeomorphisms of @xmath157 .
suppose that there exists a smooth map @xmath158 such that for any @xmath159 and any @xmath144 the equation @xmath160 is satisfied .
then , for every choice @xmath161 we have @xmath162}(z))=\det(dg_{[k]}(z))\,\mu(z),\ ] ] where @xmath163}=g_{a_k}\circ\cdots\circ g_{a_{2}}\circ g_{a_1}.$ ] _ proof .
_ it is only necessary to prove the result for @xmath90 because the general case follows easily by induction .
consider @xmath164 then @xmath165 and the lemma follows . ' '' '' _ proof of ( ii ) of theorem [ rotacions ] .
_ from part ( i ) of the theorem we know that the level sets of @xmath47 and @xmath48 in @xmath166 are diffeomorphic to circles .
moreover these functions are first integrals of @xmath44 and @xmath46 , respectively .
notice also that for any @xmath167 the lyness map @xmath168 satisfies @xmath169 with @xmath170 hence , by lemma [ mus ] , @xmath171 thus , from theorem [ teor ] , the result follows . ' '' ''
it is worth to comment that once part ( i ) of the theorem is proved it is also possible to prove that the dynamics of @xmath44 ( resp .
@xmath46 ) restricted to the level sets of @xmath47 ( resp .
@xmath48 ) is conjugated to a rotation by using that they are given by cubic curves and that the map is birational , see @xcite .
we prefer our approach because it provides a dynamical interpretation of the rotation number together with its analytic characterization .
in order to prove theorem [ main ] we need a preliminary result .
consider the maps @xmath44 and @xmath172 jointly with their corresponding first integrals @xmath47 and @xmath173 in a similar way consider @xmath174 and @xmath175 with @xmath176 and @xmath177 some simple computations prove the following elementary but useful lemma . notice that it can be interpreted as the relation between the first integrals and the non - autonomous invariants . [ elemental ] with the above notations : * @xmath178 * @xmath179 _ proof of theorem [ main ] .
_ we split the proof in two steps . for @xmath10
we first prove that there are only two types of behaviors for @xmath0 , either this set of points is formed by @xmath180 points for some positive integer @xmath181 or it has infinitely many points whose adherence is given by at most @xmath5 intervals .
secondly , in this later case , we provide an algebraic way for studying the actual number of intervals . *
first step : * we start with the case @xmath33 with the notation introduced in , it holds that @xmath182 where @xmath183 and @xmath184 .
so the odd terms of the sequence @xmath0 are contained in the projection on the @xmath185-axis of the oval of @xmath186 and the even terms in the corresponding projection of @xmath187 , where notice that we have used lemma [ elemental ] .
recall that the ovals of @xmath47 are invariant by @xmath44 and the ovals of @xmath188 are invariant by @xmath189 .
notice also that the trivial equality @xmath190 implies that the action of @xmath44 on @xmath191 is conjugated to the action of @xmath189 on @xmath192 via @xmath193 from theorem [ rotacions ] we know that @xmath44 on the corresponding oval is conjugated to a rotation of the circle .
hence , if the corresponding rotation number is rational , then the orbit starting at @xmath194 is periodic , of period say @xmath195 then the sequence @xmath0 is @xmath196-periodic . on the other hand
if the rotation number is irrational , then the orbit of @xmath194 generated by @xmath44 fulfills densely the oval of @xmath191 in @xmath36 and hence the subsequence of odd terms also fulfills densely the projection of @xmath191 in the @xmath185-axis .
clearly , the sequence of even terms do the same with the projection of the oval of @xmath197 similarly when @xmath38 the equalities @xmath198 where @xmath199 , allow to conclude that each term @xmath200 of the sequence @xmath0 where we use the notation ( [ k=3 ] ) , is contained in one of the projections on the @xmath185-axis of the ovals @xmath201 and @xmath202 and @xmath203 , according with the remainder of @xmath204 after dividing it by 3 .
the rest of the proof in this case follows as in the case @xmath33 so the first step is done .
* second step : * from the above results it is clear that the problem of knowing the number of connected components of the adherence of @xmath0 is equivalent to the control of the projections of several invariant ovals on the @xmath185-axis .
the strategy for @xmath38 , and analogously for @xmath90 , is the following .
consider the ovals contained in the level sets given by @xmath205 , @xmath202 and @xmath203 and denote by @xmath206 and @xmath207 the corresponding closed intervals of @xmath208 given by their projections on the @xmath185-axis .
we want to detect the values of @xmath209 for which two of the intervals , among @xmath210 and @xmath211 have exactly one common point .
first we seek their boundaries .
since the level sets are given by cubic curves , that are quadratic with respect the @xmath212-variable , these points will correspond with values of @xmath185 for which the discriminant of the quadratic equation with respect to @xmath212 is zero .
so , we compute @xmath213 now we have to search for relations among @xmath214 and @xmath209 for which two of these three functions have some common solution , @xmath215 these relations can be obtained by computing some suitable resultants . taking the resultants of @xmath216 and @xmath217 ; @xmath217 and @xmath218 ; and @xmath216 and @xmath218 with respect @xmath185 we obtain three polynomial equations @xmath219 , @xmath220 and @xmath221 in short , once @xmath41 and @xmath11 are fixed we have obtained three polynomials in @xmath209 such that a subset of their zeroes give the bifurcation values which separate the number of intervals of the adherence of @xmath0 .
see the results of proposition [ ex2 ] and example [ ex3 ] for concrete applications of the method . before ending the proof
we want to comment that for most values of @xmath52 , @xmath53 and @xmath222 varying @xmath209 there appear the three possibilities , namely @xmath223 , @xmath154 or @xmath155 different intervals .
the last case appears for values of @xmath209 near @xmath224 , because the first coordinates of the three points @xmath225 @xmath226 and @xmath227 almost never coincide .
the other situations can be obtained by increasing @xmath228 ' '' '' [ ex2 ] consider the recurrence with @xmath90 and @xmath2 as in taking the values @xmath229 and @xmath230 define @xmath231 where @xmath232 is the biggest positive real root of @xmath233 and @xmath234 is the smallest positive root of @xmath235 then , a. the initial condition @xmath236 gives a two periodic recurrence @xmath0 .
moreover @xmath237 b. let @xmath194 be any positive initial conditions , different from @xmath238 and set @xmath239 let @xmath240 denote the rotation number of @xmath241 restricted to the oval of @xmath242 then a. if @xmath243 with @xmath244 then the sequence @xmath0 is @xmath196-periodic .
b. if @xmath245 and @xmath246 then the adherence of the sequence @xmath0 is formed by two disjoint closed intervals . c. if @xmath245 and @xmath247 then the adherence of the sequence @xmath0 is one closed interval .
we want to remark that , from a computational point of view , the case ( i ) almost never is detected . indeed , taking @xmath52 and @xmath53 rational numbers and starting with rational initial conditions , by using mazur s theorem
it can be seen that almost never the rotation number will be rational , see the proof of ( * ? ? ?
* prop . 1 ) . therefore ,
in numeric simulations only situations ( ii ) and ( iii ) appear , and the value @xmath248 gives the boundary between them .
in general , for @xmath22 the value @xmath249 is always the root of a polynomial of degree four , which is constructed from the values of @xmath52 and @xmath53 . _
proof of proposition [ ex2 ] .
_ clearly @xmath250 is the fixed point of @xmath44 in @xmath251 some computations give the compact expression of @xmath252 to obtain the values @xmath249 we proceed as in the proof of theorem [ main ] . in general
, @xmath253 then we have to compute the resultant of the above polynomials with respect to @xmath185 .
it always decomposes as the product of two quartic polynomials in @xmath228 its expression is very large , so we only give it when @xmath229 and @xmath254 .
it writes as @xmath255 it has four real roots , two for each polynomial .
some further work proves that the one that interests us is the smallest one of @xmath256 ' '' '' we also give an example when @xmath24 but skipping all the details .
[ ex3 ] consider the recurrence with @xmath38 and @xmath2 as in taking the values @xmath257 and @xmath258 then for any positive initial conditions @xmath259 and @xmath260 , @xmath261 moreover if the rotation number of @xmath46 associated to the oval @xmath205 is irrational then the adherence of @xmath0 is given by : * three intervals when @xmath262 where @xmath263 * two intervals when @xmath264 where @xmath265 * one interval when @xmath266 the values @xmath249 and @xmath267 are roots of two polynomials of degree 8 with integer coefficients that can be explicitly given .
from theorem [ rotacions ] it is natural to introduce the _ rotation number function _ for @xmath44 : @xmath268 where @xmath269 , as the map that associates to each invariant oval @xmath191 , the rotation number @xmath54 of the function @xmath44 restricted to it .
the following properties hold : a. the function @xmath54 is analytic for @xmath27 , @xmath270 and it is continuous at @xmath271 this can be proved from the tools introduced in ( * ? ? ?
4 ) . b. the value @xmath272 is given by the argument over @xmath273 of the eigenvalues ( which have modulus one due to the integrability of @xmath44 ) of the differential of @xmath44 at @xmath123 . c. @xmath274 d. @xmath275 where @xmath276 is the rotation number and @xmath277 when one considers its rotation number .
so , while for us the rotation number of the classical lyness map for @xmath278 is @xmath279 , in other papers it is computed as @xmath280 function associated to the classical lyness map .
then , from the results of @xcite we know that @xmath281 that for @xmath282 positive , @xmath283 is monotonous and @xmath284 note that item ( iii ) follows because @xmath189 is conjugated with @xmath44 via @xmath285 which is a diffeomorphism of @xmath36 , because @xmath286 since @xmath287 preserves the orientation , the rotation number functions of @xmath189 and @xmath44 restricted to the corresponding ovals must coincide .
we are convinced that when @xmath289 and @xmath290 @xmath291 but we have not been able to prove these equalities . if they were true , by combining them with the values of the rotation number function at @xmath292 this would give a very useful information to decide if , apart from the trivial cases @xmath293 there are other cases for which the rotation number function is constant .
notice that in these situations the maps @xmath44 or @xmath46 would be globally periodic in @xmath251 this information , together with the values at @xmath294 , also would be useful to know the regions where the corresponding functions could be increasing or decreasing .
finally notice that this rotation number at infinity is not continuous when we approach to @xmath295 or @xmath296 , where the recurrence and the first integral are also well defined on @xmath251 for instance @xmath297 and the numerical experiments of next subsection seem to indicate that for @xmath289 or @xmath290 @xmath298 before proving theorem [ teonomonotonia ] we introduce with an example the algorithm that we will use along this section to compute lower and upper bounds for the rotation number . we have implemented it in an algebraic manipulator .
notice also that when we apply it taking rational values of @xmath52 and @xmath53 and rational initial conditions , it can be used as a method to achieve proofs , see next example or the proof of theorem [ teonomonotonia ] .
fix @xmath299 @xmath300 and @xmath301 then @xmath302 compute for instance the 27 points of the orbit starting at @xmath303 @xmath304 and consider them as points on the oval @xmath305 see figure [ fig1 ] .
we already know that the restriction of @xmath306 to the given oval is conjugated to a rotation , with rotation number @xmath310 that we want to estimate .
this can be done by counting the number of turns that give the points @xmath311 , after fixing some orientation in the closed curve .
we orientate the curve with the counterclockwise sense .
so , for instance we know that the second point @xmath312 has given more that one turn and less than two , giving that @xmath313 and hence that @xmath314 doing the same reasoning with all the points computed we obtain , @xmath315 where we have only written the more relevant informacions obtained , which are given by the points of the orbit closer to the initial condition .
so , we have shown that @xmath316 in figure [ fig2 ] we represent several successive lower and upper approximations obtained while the orbit is turning around the oval .
we plot around six hundred steps , after skipping the first fifty ones . by taking 1000 points
we get @xmath317 and after 3000 points , @xmath318 in fact when we say that @xmath319 , the value @xmath320 is the upper lower bound obtained by following all the considered points of the orbit , and @xmath321 is the lowest upper bound .
notice that taking 1000 or 3000 points we have obtained the same lower bound for @xmath322 _ proof of theorem [ teonomonotonia ] .
_ consider @xmath323 @xmath324 and the three points @xmath325 notice that @xmath326 hence @xmath327 by applying the algorithm described above , using 100 points of each orbit starting at each @xmath328 we obtain that @xmath329 since @xmath330 we have proved that the function @xmath331 has at least a local maximum in @xmath332 from the continuity of the rotation number function , with respect @xmath41 and @xmath209 , we notice that this result also holds for all values of @xmath52 and @xmath53 in a neighborhood of @xmath333 we start by studying with more detail the rotation number function @xmath331 , that we have considered to prove theorem [ teonomonotonia ] . in this case
the fixed point is @xmath335 and @xmath336 moreover @xmath337 . by applying our algorithm for approximating the rotation number , with @xmath338 points on each orbit
, we obtain the results presented in table 1 . in figure [ fig3 ]
we also plot the upper and lower bounds of @xmath331 that we have obtained by using a wide range of values of @xmath228 . on the horizontal axis we represent @xmath342 and , on the vertical axis , the value of the rotation number .
notice that for values of @xmath342 smaller that 70 both values are indistinguishable in the figure.,width=14,height=7 ] for other values of @xmath52 and @xmath53 we obtain different behaviors .
all the experiments are performed by starting at the fixed point @xmath343 and increasing the energy level by taking initial conditions of the form @xmath344 , by decreasing @xmath185 to @xmath307 . with this process
we take orbits approaching to the boundary of @xmath36 , that is lying on level sets of @xmath47 with increasing energy .
the step in the decrease of @xmath185 ( and therefore in the increase of @xmath209 ) is not uniform , and it has been manually tuned making it smaller in those regions where a possible non monotonous behavior could appear .
consider the set of parameters @xmath345 , where notice that we also consider the boundaries @xmath295 or @xmath346 where the map @xmath44 is well defined .
we already know that the rotation number function behaves equal at @xmath347 and @xmath348 moreover we know perfectly its behavior on the diagonal @xmath349 ( when @xmath350 it is monotonous decreasing and when @xmath351 it is monotonous increasing ) and that @xmath352 and @xmath353 hence a good strategy for an numerical exploration can be to produce sequences of experiments using our algorithm by fixing some @xmath354 and varying @xmath53 .
for instance we obtain : * case @xmath355 for all the values of @xmath356 considered , the rotation number function seems to tend to @xmath357 when @xmath209 goes to infinity .
moreover it seems * * monotone decreasing for @xmath358 * * to have a unique maximum when @xmath359 ; * * monotone increasing for @xmath360 * case @xmath361 for all the values of @xmath356 considered , the rotation number function seems to tend to @xmath362 when @xmath209 goes to infinity .
moreover it seems * * monotone decreasing for @xmath363 ; * * to have a unique maximum when @xmath364 ; * * monotone decreasing for @xmath365 .
the above results , together with some other experiments for other values of @xmath52 and @xmath53 , not detailed in this paper , indicate the existence of a subset of positive measure in @xmath366 where the corresponding rotation number functions seem to present an unique maximum .
this subset probably separates two other subsets of @xmath366 , one where @xmath54 is monotonically decreasing to @xmath357 , and another one where @xmath54 increases monotonically to the same value .
the `` oscillatory subset '' seems to shrink to @xmath367 when it approaches to the line @xmath368 and seems to finish in one interval on each of the borders @xmath369 and @xmath370 .
further analysis must be done in this direction in order to have a more accurate knowledge of the bifurcation diagram associated to the behavior of @xmath371 on @xmath366 . | we describe the sequences @xmath0 given by the non - autonomous second order lyness difference equations @xmath1 where @xmath2 is either a 2-periodic or a 3-periodic sequence of positive values and the initial conditions @xmath3 are as well positive .
we also show an interesting phenomenon of the discrete dynamical systems associated to some of these difference equations : the existence of one oscillation of their associated rotation number functions .
this behavior does not appear for the autonomous lyness difference equations . _
2000 mathematics subject classification : _ ` 39a20 , 39a11 ` _ keywords : _ difference equations with periodic coefficients , circle maps , rotation number . |
bl lacertae objects ( bl lacs ) are members of the blazar class of active galactic nuclei ( agn ) .
blazars exhibit rapid , large amplitude variability at all wavelengths , high optical and radio polarization , apparent superluminal motion , and in some cases , gamma - ray emission .
all of these observational properties lead to the broadly held belief that blazars are agn with jets oriented nearly along our line of sight .
the broadband spectral energy distribution of blazars , when plotted as @xmath8f@xmath9 versus frequency , shows a double peaked shape , with a smooth extension from radio to between ir and x - ray frequencies ( depending on the specific blazar type ) , followed by a distribution that typically starts in the x - ray band and can peak in the gamma - ray band , at energies as high as several hundred gev .
the low energy part is believed to be incoherent synchrotron radiation from a relativistic electron - positron plasma in the blazar jet .
the origin of the high energy emission is still a matter of considerable debate ( e.g. , buckley 1998 ; mannheim 1998 ) .
markarian 501 ( mrk 501 ) is one of the closest bl lacs ( @xmath10 ) known and it is one of the brightest in the x - ray band . because its peak spectral power output occurs at uv / x - ray energies , mrk 501 is classified as a high frequency peaked bl lac .
it is one of only six bl lacs reported as sources of very high energy ( vhe , e @xmath11 250 gev ) gamma rays and , along with mrk 421 , one of only two that have been confirmed as vhe sources ( for a review , see catanese & weekes 1999 ) .
mrk 501 has also been detected as a source of gamma rays at gev energies with the energy gamma - ray experiment telescope ( egret ; kataoka et al .
1999 ) and at a few hundred kev by the oriented scintillation spectrometer experiment ( osse ; catanese et al .
1997 ) on the _ compton gamma - ray observatory_. just as all other blazars , mrk 501 exhibits rapid , large amplitude variability over a wide range of wavelengths . in x - rays , variations of from 30% to 300% were observed on time scales of days in 1997 during a high emission state @xcite but no sub - day scale flares were seen .
the fastest variation observed in x - rays was a flux increase of approximately 20% in about 12 hours seen with exosat in 1986 @xcite .
spectral variability in x - rays from mrk 501 has been both moderate , with changes in the spectral index of @xmath00.1 - 0.3 on several day scales ( e.g. , pian et al .
1998 ) , and rapid , with spectral variations of @xmath00.5 on 2 - 3 day time - scales , observed in 1998 june @xcite . the whipple observatory has observed vhe gamma - ray variations spanning a factor of @xmath1170 in flux in four years of observations and has observed significant variability with time - scales as short as 2 hours @xcite .
similar variability ranges are observed by other vhe telescopes @xcite . in the r - band , @xcite reported the detection of a flare in which the flux increased by 4% ( from @xmath12 to 13.80 ) in 15 minutes with a decay to the previous level in the same amount of time .
multi - wavelength observations have revealed correlations between vhe gamma rays and x - rays in this object @xcite and in 1997 , the synchrotron spectrum of mrk 501 was observed to extend up to approximately 100 kev @xcite , the highest seen in any blazar and a 50-fold increase over what was observed only one year before @xcite .
this behavior has established mrk 501 as the prototype for a subset of bl lacs that exhibit large shifts in the peak of their synchrotron spectra during flares . in this paper , we report on observations of mrk 501 taken with the _ rossi x - ray timing explorer _ ( _ rxte _ ) in 1998 may as part of a multi - wavelength campaign .
the full multi - wavelength results will be reported in a future work . here
, we concentrate on the observation of a very rapid flare and discuss its implications .
_ rxte _ consists of the proportional counter array ( pca ; jahoda et al .
1996 ) which is sensitive to 2 - 60 kev photons , the high energy x - ray transient experiment ( hexte ; rothschild et al .
1998 ) which is sensitive to 15 - 150 kev photons , and the all - sky monitor ( asm ; levine et al . 1996 ) which is sensitive to 2 - 12 kev photons . here
we report on the results of the pca observations .
the count rate for hexte was too low to obtain significant count rate variations on the time scales observed .
there are no observations with the asm during the period of the flare . also , because the x - ray flare occurred at approximately 18 hours universal coordinated time ( utc ) , there are no overlapping tev observations from whipple , hegra , cat , or the telescope array . the observations of mrk 501 occurred between 1998 may 15 and 29 .
they consisted of roughly four 2 - 3 ks pointings per day for the full two weeks .
after screening for good time intervals , as described below , the data set consists of 114 ks of observations .
each observation resulted in a very significant detection of mrk 501 that allowed spectra to be resolved and short - term variability to be investigated .
we used ftools v4.2 to analyze this data .
the background was estimated using the weak source background models ( appropriate for sources with count rates @xmath140 counts / s / pcu ) and the latest response matrices obtained from the _ rxte _ guest observer facility ( gof ) web site.1st.html ] good time intervals were selected from the standard 2 data files using the screening criteria recommended by the _ rxte _
finally , only proportional counter units ( pcus ) 0 , 1 , and 2 were active for the vast majority of the observations presented here , so we only use those pcus in this work .
for spectra and all other light curves , we used all three layers .
spectral fits were performed using xspec 10.0 .
we use the galactic hydrogen column density of @xmath13 @xmath6 @xcite to model the effect of photoelectric absorption , which is negligible for the energy range covered by the pca for this object located far from the galactic plane .
the 2 - 10 kev lightcurve for the 1998 may observations is shown in figure [ 1998_lc ] .
the data points are shown in 1024-second bins .
the count rate varies by approximately a factor of 1.8 from 21 cts / s to 38 cts / s . by comparison ,
_ rxte _ observations of mrk 501 in 1997 april exhibited 2 - 10 kev lightcurve count rates of 80 cts / s to 160 cts / s @xcite .
thus , mrk 501 was in a much lower emission state during these observations than in 1997 , though the flux was relatively high by historical standards .
the light curve begins with a 20% drop over two days , follows with a 50% rise over three days , and ends with a gradual decline in flux of 65% over the remaining nine days of the observation . during the decline , small amplitude ,
day - scale flares are evident at mjd 50955 and 50960 .
most notably , toward the end of mjd 50958 ( indicated by the two vertical lines in fig .
[ 1998_lc ] ) , the flux is much higher than the surrounding observations , and is actually the highest count rate in this light curve . for the remainder of the paper we concentrate on this flare observation .
detailed analysis of the entire 1998 data set will be presented in a future paper .
this flare occurs during what is otherwise a somewhat unremarkable observation period .
there is no evidence of significant short term variability within any other single observation nor is the variability seen on day scales in the other observations of such large amplitude .
the largest variation on any other day is a 30% increase of the flux between mjd 50951 and 50952 .
our comparisons of background files with the pca background model and investigations of the data quality monitors ( e.g. , electron activation ) confirm that this flare is not a spurious effect . in figure [ flare_lc ] , we show the 2 - 10 kev and 10 - 15 kev light curves for the observation indicated by the vertical lines in figure [ 1998_lc ] . the observation was taken on may 25 between 17.9 and 18.6 hours universal coordinated time ( utc ) .
the data are shown in 96-second bins . during this observation ,
mrk 501 exhibited low flux , very steady emission for about half of the observation .
after this , the count rate increased from approximately 26 cts / s to 41 cts / s in approximately 200 seconds , corresponding to a brightness increase of 13%/minute .
this is followed by a steady decrease in the count rate to approximately 30 cts / s in approximately 580 seconds , a decline rate of 3%/minute .
though not quite as well measured due to lower statistics , the flare is clearly evident in the 10 - 15 kev light curve , with a similar rise and fall time - scale .
the 10 - 15 kev flux variation rate is 18%/minute and 3%/minute for the rise and fall of the flare , respectively .
observations were also taken with _ rxte _
approximately 5.5 hours before and 5.5 hours after this observation .
neither showed any significant variability .
the observation before the flare had an average 2 - 10 kev count rate of 24.5 cts / s and the one after the flare had an average count rate of 27 cts / s .
both are consistent with little or no change in flux compared to that seen at the start and end of the flare observation .
the average spectrum for the observations between may 15 and 29 is best fit by a broken power law model with a photon spectral index of [email protected] up to a break at [email protected] kev , above which the spectral index is [email protected] .
the 2 - 10 kev flux during this period is @xmath14 erg @xmath6 s@xmath7 .
the spectrum extends to at least 40 kev with no evidence of another break .
this spectrum connects smoothly with osse measurements during this period @xcite .
observations taken with _ beppo sax _ on april 28 and 29 and may 1 @xcite indicate a somewhat harder spectrum with peak power output at @xmath020 kev .
the x - ray flux during the _ beppo sax _
observation is @xmath1550% higher than in the _ rxte _ observations reported here , so the spectral shift is consistent with the previously observed tendency of mrk 501 to increase the energy at which the spectral energy distribution peaks as its flux increases ( e.g. , pian et al .
1998 ) .
spectral analysis of the flare observation reveals a rapid change during the course of the flare .
we fit the spectrum from 3 - 15 kev where there are sufficient photon statistics and no problems with the pca response function .
the average spectrum for the observation is well - fit by a simple power law with a photon spectral index of @xmath16 . to investigate the evolution of the spectrum during the course of the flare , we break up the data into three regions : before the flare , during the rise , and during the decay .
all are well - fit by simple power laws and a summary of those fits is given in table [ spec_fits ] .
the photon spectral index is @xmath17 before the flare , @xmath18 during the rise of the flare , and @xmath19 during the decay of the flare .
the spectrum during the flare follows this power law at least out to 30 kev , indicating a large shift in the location of the peak power output during the flare .
dividing the decay of the flare into two parts does not reveal significantly different spectra than the average spectrum for the entire decay region .
the observation taken approximately 5.5 hours before the flare indicates a spectral index of @xmath20 and a 2 - 10 kev flux of @xmath21 erg @xmath6 s@xmath7 , consistent with the observations just prior to the flare .
the observation taken approximately 5.5 hours after the flare observation indicates a spectral index of @xmath22 and a 2 - 10 kev flux of @xmath23 erg @xmath6 s@xmath7 , indicating a significant softening of the spectrum following the flare .
the variation in the spectral index during the course of the flare can provide insights into the dominant flaring timescales and acceleration process . as discussed by @xcite , for a flare in which the variability and acceleration time - scales are much less than the cooling time - scale a plot of the spectral index versus flux should follow a clockwise pattern , i.e. , the harder energies vary first . for a flare
where the variability , acceleration , and cooling time - scales are similar , the spectral index versus flux diagram should move in a counter - clockwise direction , i.e. , the softer energies vary first because the number of particles changes due to the acceleration process which proceeds from low energy to high energy .
clockwise patterns are most commonly observed in the tev sources mrk 421 ( e.g. , takahashi et al . 1996 ) and pks 2155 - 304 ( e.g. , kataoka et al . 2000 ) but counter - clockwise patterns have been recently observed from these objects @xcite . because the data do not have sufficient statistics to plot spectral index versus flux on such short time scales , we instead plot the hardness ratio ( 10 - 15 kev count rate/2 - 10 kev count rate ) versus flux for the flare on may 25 in figure [ hard_flux ] .
the numbers in the plot indicate the development of the hardness ratio in time during the flare .
the large cluster of filled circles represents the observations before the onset of the flare .
the filled triangles represent the rising part of the flare . the point indicated by the tail of the arrow marked with
the `` 1 '' is the last low flux point before the flare starts .
the filled squares are data taken during the decay of the flare . during the rise of the flare , the hardness ratio increases steadily . during the decay of the flare
, there is a slight trend for the hardness ratio to increase even further .
thus , these observations are consistent with a counter - clockwise pattern .
a clockwise pattern seems precluded by the significantly harder spectrum during the decay of the flare than the rise ( see table [ spec_fits ] ) but other patterns in the hardness ratio versus flux diagram can not be ruled out given the statistical errors in this observation . the counter - clockwise pattern and the large shift in the synchrotron peak
imply that the acceleration process dominates the development of the flare , accelerating a fresh population of high energy electrons which causes the flare . in summary ,
_ rxte _ observations of mrk 501 in 1998 may have revealed a flux state which was approximately one - fourth as strong as observed in 1997 , and an average spectrum with peak power output at approximately 6 kev .
this is a decrease of more than a factor of 15 from the 100 kev peak seen in 1997 . during these observations
, a very rapid flare was observed in which the location of peak power output increased from @xmath33 kev to @xmath2430 kev .
this large shift in peak power output energy is similar to the behavior of mrk 501 in 1997 and in 1998 june .
the evolution of the hardness ratio of the flare is consistent with the flare development being dominated by the acceleration process but the lack of simultaneous multi - wavelength observations prohibits further detailed testing of emission models .
though rapid variations have been seen from mrk 501 at other wavelengths before , it was generally regarded as a more slowly varying object than the other tev sources , mrk 421 and pks 2155 - 304 .
thus , one could conduct less dense observations of mrk 501 and still sample the variations with adequate coverage to resolve the shape of the variations and the correlations between wavelengths .
these observations show that very dense multi - wavelength observations are required for mrk 501 as well since it can vary on time - scales comparable to the fastest seen in the other tev sources .
they also indicate that , as has been seen in mrk 421 @xcite , these very rapid flares can occur when the source is not in a particularly high emission state , so dense observations must be used regardless of the flux level observed from these objects .
multi - wavelength observations of such rapid flares will provide stringent tests of emission models for these tev sources and may lead to a better understanding of the acceleration process that occurs in their jets .
the authors wish to thank k. jahoda and t. jaffe for advice about the _ rxte _ analysis .
mc wishes to thank t. weekes , d. carter - lewis , j. finley , j. buckley , c. dermer , n. johnson , and f. krennrich for their support of these observations .
mc acknowledges grant support from nasa and the u. s. department of energy .
rms acknowledges support from nasa contract nas38252 .
jahoda , k. , swank , j. h. , giles , a. b. , stark , m. j. , strohmayer , t. , zhang , w. , & morgan , e. h. 1996 , in `` euv , x - ray , and gamma - ray instrumentation for astronomy vii , '' spie proceedings , ed .
o. siegmund & m. gummin , 2808 , 59 sambruna , r. m. 2000 , in gev - tev astrophysics : towards a major atmospheric cherenkov telescope vi , ed
. b. dingus , m. h. salamon , & d. kieda ( new york : aip ) , aip conf .
, in press ( astro - ph/9912129 ) cccc 17.858 18.524 & [email protected] & [email protected] & 0.895 ( 30 ) + 17.858 18.284 & [email protected] & [email protected] & 0.768 ( 30 ) + 18.284 18.311 & [email protected] & [email protected] & 0.483 ( 30 ) + 18.311 18.524 & [email protected] & [email protected] & 0.571 ( 28 ) + | we present x - ray observations of the bl lacertae ( bl lac ) object markarian 501 ( mrk501 ) , taken with the _ rossi x - ray timing explorer _ in 1998 may as part of a multi - wavelength campaign .
the x - ray light curve shows a very rapid flare in which the 2 - 10 kev flux increased by @xmath060% in @xmath1200 seconds .
this rapid rise is followed by a drop - off in the 2 - 10 kev flux of @xmath040% in @xmath1600 seconds .
the 10 - 15 kev variation in this flare is roughly a factor of two on similar time - scales . during the rise of the flare , the 3 - 15 kev spectral index hardened from [email protected] to [email protected] , where it remained during the decay of the flare
this is the fastest variation ever seen in x - rays from mrk501 and among the fastest seen at any wavelength for this object .
the shift in the energy at which the spectral power peaks ( from @xmath33 kev before the flare to @xmath430 kev during the flare ) is also among the most rapid shifts seen from this object .
this flare occurs during an emission state ( 2 - 10 kev flux @xmath5 erg @xmath6 s@xmath7 ) that is approximately 25% of the peak flux observed in 1997 april from this object but which is still high compared to its historical average x - ray flux .
the variations in the hardness ratio are consistent with the low energy variations leading those at high energies during the development and decay of the flare .
this pattern is rare among high frequency peaked bl lac objects like mrk 501 , but has been seen recently in two other tev emitting bl lacs , mrk 421 and pks 2155 - 304 .
the hard lag is consistent with a flare dominated by the acceleration time - scale for a simple relativistic shock model of flaring . |
typical results for @xmath40 _ vs. _ @xmath16 are shown in fig .
[ fig1 ] , for both @xmath41 and @xmath42 .
the low field portion is further shown in detail in the inset , which reveals essentially no change until a critical field @xmath24 is reached , above which @xmath43 increases rapidly .
this critical field @xmath24 represents the field at which flux first enters the sample and is governed by both the meissner state of the superconductor and an effective geometric barrier at the surface .
we are justified in calling it the lower critical field as demagnetization corrections are negligible in the case where @xmath44 .
note that the presence of surface barriers would lead to hysteresis in the field dependence of @xmath45 at the onset of flux entry . in our data ,
the observed hysteresis is minimal and is within the experimental resolution .
it is also to be noted that surface barrier effects would scale the field values but do not affect the temperature dependence .
0.45 notice that unlike magnetization measurements , where @xmath24 is deduced from deviations from linearity in the @xmath46 curve , the signature in our data at @xmath24 is very sharp .
this is because the experiment effectively measures the flux density @xmath47 ( as will become evident later ) .
hence , the signature at @xmath24 is a change from zero , and not a change of slope as in @xmath46 data .
0.45 the @xmath2 dependence of @xmath48 for @xmath13 is shown in fig .
this essentially has a linear behaviour from @xmath49 to @xmath50 , well represented by @xmath51 , where @xmath52 .
this is unlike that seen in conventional low @xmath49 superconductors and even differs from the behavior of @xmath53 .
however this anomalous behavior finds support in measurements on a related quantity , the superfluid density , measured via the condensate contribution to the complex conductivity , @xmath54 , which we have measured elsewhere @xcite using a @xmath55 @xmath56 cavity .
there , @xmath54 was shown to have an anomalous temperature dependence well represented by @xmath57 ( also shown in fig .
[ fig2 ] ) .
now @xmath58 , and since @xmath59 $ ] , there is a good correlation between the linear temperature dependence of the @xmath24 and @xmath60 measurements .
( note that for clean type - ii superconductors , @xmath61 varies as @xmath62 and thus @xmath63 is expected to have a weak @xmath2 dependence@xcite ) .
above @xmath24 , @xmath64 increases strongly in the @xmath65 case , due to the increasing @xmath66 in the sample .
a similar increase also occurs in the @xmath67 data , although here the effective @xmath68 due to demagnetization .
this field dependence @xmath69 for @xmath70 arises from two effects of the magnetic field on the superconductor : depairing of condensed electrons ( which dominates when the rf current @xmath71 ) and dynamic vortex response ( which dominates when @xmath72 ) .
this is followed by a sharp increase starting and terminating at field values which we call @xmath6 and @xmath73 respectively . above @xmath7 ,
@xmath19 increases almost linearly with @xmath16 , which is consistent with the normal state showing a strong positive magnetoresistance .
there is an additional field @xmath8 just above @xmath7 which can be identified as the point where the @xmath74 and @xmath42 curves meet as seen in fig . 1 .
we have observed a similarly rich structure in the transition region of the @xmath75 and @xmath76 compounds @xcite .
the presence of multiple structure in the transition region , although interesting in its own right , makes the identification of a single upper critical field @xmath77 somewhat ambiguous .
this difficulty is also compounded by the linear-@xmath16 magnetoresistance in the normal state .
all three field scales ( @xmath6 , @xmath7 and @xmath8 ) agree quantitatively with the @xmath77 values reported in the literature@xcite .
while the exact origin of this structure is not yet precisely known , its presence in all the borocarbide superconductors we have studied seems to suggest an intrinsic phenomenon and is not likely to be due to spurious chemical phases , as will be presented in a forthcoming publication @xcite . in this paper , we focus on the general nature of the depairing and vortex contributions to the rf penetration depth . to test the validity of the standard coffey - clem model for the depairing and vortex response in this system , an effective upper critical field @xmath77
coinciding with @xmath7 is assumed for subsequent analysis .
it is to be pointed out that our choice of @xmath7 as the effective upper critical field is not arbitrary and is based on other factors like the validity of the bardeen - stephen approximation @xcite which relates the upper critical field to the flux flow resistivity .
a discussion on flux viscosity presented later in the paper will clarify this issue further .
we now turn to an understanding of the field dependence of @xmath78 .
the principal causes for this field dependence are the depairing effect of @xmath47 on the condensate and the dynamic response of the vortices to the lorentz force induced by the rf current .
these contributions have been treated self - consistently by coffey and clem , leading to the following expression for the complex penetration depth which includes the full angular dependence : @xcite @xmath79 ^{1/2 } \label{eq : cc}\ ] ] where @xmath80 and @xmath81 are the condensate penetration depth and the normal - fluid skin depth in the superconducting state , and @xmath82 and @xmath83 are the vortex pinning force constant and viscosity respectively .
the field - dependence of the condensate background is represented by @xmath84 and @xmath85 $ ] , where for instance @xmath86\left [ 1-(b / b_{c2})\right ] $ ] in a two - fluid model . for @xmath41 ,
the rf currents in the @xmath17 plane are parallel to the vortices and hence the lorentz force is zero for those vortices at the sample faces .
( a schematic representation of the relative orientations of @xmath87 and the rf current , @xmath88 , with respect to @xmath16 is shown as an inset in fig .
[ fig1 ] ) .
thus the vortex response is mostly absent in the @xmath41 case , except for a small contribution of order @xmath89 from the edges , where @xmath90 and @xmath91 are the sample thickness and width respectively .
this can be incorporated phenomenologically in terms of an aspect ratio @xmath92 ^{-1}$ ] which mixes the vortex contribution ( @xmath67 ) to the force - free ( @xmath41 ) term as @xmath93 .
the angular dependence provides a nice way of separating out the depairing and the vortex contributions to the penetration depth .
0.45 in fig .
[ fig : norm ] we present the same data as in fig .
[ fig1 ] normalized with respect to @xmath94 . motivated by the observation that the field dependence of @xmath95 follows that of @xmath96 we have used @xmath97 for this normalization .
theoretical calculations based on eq . [ eq : cc ] , taking into account the small aspect ratio as discussed earlier and also assuming negligible pinning ( i.e. @xmath98 @xmath99 ) , are shown in fig .
[ fig : norm ] along with the data . for the vortex contribution , at fixed frequency
@xmath100 the response is pinning dominated if @xmath101 and viscosity dominated if @xmath102 , where @xmath103 .
it is quite striking that pinning is negligible even at @xmath15 , which would effectively imply that flux is in free flow as soon as vortices enter the sample above @xmath24 .
this is in contrast to the estimated @xmath104 values of @xmath105 in @xmath106@xcite and @xmath107 in low @xmath49 materials @xcite . for @xmath41 , the low field variation of @xmath43
is adequately reproduced by the calculation , confirming the presence of a finite vortex contribution consistent with the aspect ratio independently determined from the sample dimensions .
the @xmath67 data shows a dip at a field below the effective @xmath108 .
this can result from sudden changes in pinning or scattering which are parametrized by the terms @xmath98 and @xmath109 in eq .
[ eq : cc ] . a rise in @xmath98 at a threshold field
can be associated with the `` peak effect '' commonly seen in type - ii superconductors with weak pinning .
the signature of the peak effect is a peak in critical current density and would show up as a dip in the rf susceptibility , which is what we are effectively measuring in the @xmath67 case .
there is a general consensus that the peak effect arises due to enhanced pinning as a result of a rapidly decreasing rigidity of the vortex lattice just below @xmath110@xcite .
however , there is still some controversy over whether it is a softening of the tilt modulus ( @xmath111 ) or shear modulus ( @xmath112 ) that leads to this decrease in rigidity @xcite .
an alternate possibility is that this dip feature is due to an anomalous field dependence of @xmath109 arising as a result of unusual quasiparticle scattering in the superconducting state .
since the effect is much stronger in the vortex response rather than the depairing , this is an indication that the scattering due to bound quasiparticle states in the normal vortex cores could also be responsible for the observed dip in the @xmath67 data below the effective @xmath77 .
when the vortex term dominates as at @xmath67 , the @xmath113 data can be used to extract the pinning forces and viscosity as we have done elsewhere in both single crystals and films of @xmath114@xcite .
a bardeen - stephen estimate of the viscosity can be made using @xmath115 , which yields @xmath116 for @xmath13 , compared to values of @xmath117 for @xmath56 .
0.45 the data in fig .
[ fig : norm ] can also be presented as the normalized flux flow resistivity @xmath118 which is presented as a function of @xmath16 for @xmath119 in fig .
[ fig : ve ] .
the flux flow resistivity approaches the normal state almost linearly with a slope @xmath120 @xmath121 consistent with the observations of gittleman and rosenblum in conventional superconductors@xcite .
overall the response appears to be well described by eq .
[ eq : cc ] assuming @xmath122 and @xmath123 , consistent with magnetization experiments which have also shown evidence for extremely weak pinning@xcite .
the differences between the observed response and theoretical bardeen - stephen response ( indicated by the solid line in fig .
[ fig : ve ] ) can be attributed to an underestimate of @xmath124 in the superconducting state as well as some variation of @xmath82 .
if the dip observed in the @xmath67 data is associated with a change in pinning , we can estimate the change in @xmath98 in the peak region from the relative depth of the dip since @xmath125 .
a plot of @xmath126 extracted in this form is shown in the inset to fig .
[ fig : ve ] .
0.45 the full angular dependence of @xmath43 for @xmath127 is shown in fig . [
fig : ang ] along with the fit generated using eq .
[ eq : cc ] .
the inset shows @xmath128 plotted against @xmath129 for the same data . a similar plot for a poly - crystalline @xmath56 sample of comparable dimensions is also shown in the panel .
the excellent agreement between the experimental data and the coffey - clem expression of eq .
[ eq : cc ] is obvious from the good fit and also the linear dependence of @xmath128 with @xmath129 .
the slopes are vastly different for @xmath13 and @xmath56 as the response is pinning dominated in @xmath56 whereas pinning is almost irrelevant in @xmath13 at @xmath15 .
the coffey - clem calculation as presented in eq .
[ eq : cc ] assumes an isotropic @xmath77 case and from the good fit with the @xmath13 data one can infer that the effective @xmath77 is generally isotropic in this system@xcite , although a slight deviation of the data from the model at low angles might indicate the presence of some degree of anisotropy in these layered systems .
elsewhere we have observed unusual angular dependence and anisotropy effects in @xmath130 related to magnetic order and details about this will be given in a separate publication .
the magnitude and @xmath2 dependence of the three features in the vicinity of @xmath108 for @xmath13 ( @xmath6 , @xmath7 and @xmath8 ) are shown in fig .
[ fig : ht ] are in good agreement with those determined by magnetization measurements @xcite and the overall temperature dependence is fit reasonably well by a @xmath131 form .
two aspects make the observed onset field @xmath6 extremely interesting : @xmath6 is located at about the same field where the dip is seen in the @xmath132 data and occurs further from the effective upper critical field than would otherwise be expected .
the theoretical calculation fails to reproduce the finite slope of the transition , which would imply an unusual field dependence of @xmath133 .
looking at it in a different way , the simple variation of the superfluid density ( @xmath134 ) as ( @xmath135 ) within the framework of a two - fluid model breaks down and one has to assume a stronger reduction in the superfluid density as @xmath136 . note that earlier studies on @xmath13@xcite using a superconducting @xmath56 cavity at @xmath55 also indicated disagreement with bcs or two - fluid models in the @xmath2 dependence of the surface resistance and penetration depth , as displayed in fig .
[ fig2 ] in terms of @xmath137 .
this may indicate the existence of scattering of unknown origin in the superconducting state .
we believe the origin of @xmath138 could result from this anomalous scattering or vortex lattice softening / melting or a combination of both these effects . whatever the cause , the underlying dependence of @xmath139 in the vicinity of @xmath6 is masked to some extent by the mixing of vortex and depairing terms in the @xmath140 data . in conclusion , the rf measurements presented here reveal a rich variety of electrodynamic response in the superconducting state of the non - magnetic borocarbide superconductor @xmath13 .
in particular , @xmath141 and @xmath24 both have anomalous non bcs @xmath2 dependences .
the experiments have very high sensitivity which enable observation of subtle effects on the superconducting order parameter and the vortex lattice due to the applied field .
a unique way of studying the vortex response and field - induced depairing individually by variation of @xmath18 is presented .
the angular dependence of @xmath142 is well described by the coffey - clem theory with an isotropic @xmath77 .
a dip is observed in the vortex response which can be related to a change in pinning ( and thus to the `` peak effect '' ) or alternatively to anomalous scattering in the mixed state .
pinning is found to be extremely weak , and consequently , the crossover frequency is expected to be orders of magnitude lower than that of most conventional and high-@xmath49 superconductors .
structure is observed in the depairing transition from the superconducting to normal state , which is also found to have an anomalous @xmath16 dependence .
thus although @xmath13 is in many ways a conventional high-@xmath143 type - ii superconductor , the high frequency measurements appear to indicate many unconventional properties of the meissner and mixed states .
work at northeastern was supported by grant nsf - dmr-9223850 .
ames laboratory is operated for the u. s. department of energy by iowa state university under contract no .
work at ames was supported by the director for energy research , office of basic energy sciences .
r. j. cava , h. takagi , b. battlog , h. w. zandbergen , j. j. krajewski , w. f. p. jr .
, r. b. van dover , r. f. felder , t. siegrist , k. mizuhaski , j. o. lee , h. eisaki , s. a. carter , and s. uchida , nature * 367 * , 146 ( 1994 ) . | measurements of the rf penetration depth @xmath0 are used to study the superconducting order parameter , vortex dynamics in the mixed state and delineate critical fields in the borocarbide superconductor @xmath1 .
the lower critical field has an anomalous @xmath2 dependence , @xmath3 { \rm ~koe}$ ] , which is however consistent with independent superfluid density measurements at microwave frequencies .
the vortex response is dominated by viscous flux flow , indicative of extremely weak pinning , and is parametrized by a field scale @xmath4 .
the angular dependence of the vortex contribution @xmath5 is in good agreement with the coffey - clem model .
structure is seen in the depairing transition in the vicinity of the upper critical field , with the existence of well - defined critical fields @xmath6 , @xmath7 and @xmath8 , with the vortex field scale @xmath9 closest to @xmath7 .
overall the measurements indicate that @xmath10 has a rich and unusual field dependence of its transport parameters . 2 the borocarbide superconductors lnni@xmath11b@xmath11c where ln = \{y , lu , tm , er , ho and dy } @xcite are a recently discovered family of superconductors with elevated transition temperatures . while there have been numerous studies of these superconductors using traditional probes such as magnetization , specific heat , tunneling , etc .
there have been relatively few which directly probe the order parameter and dynamics of vortices in this paper we report on studies of the radio frequency ( rf ) penetration depth @xmath12 of single crystal @xmath13 ( @xmath14 ) which were carried out in a highly sensitive rf tunnel diode oscillator set - up .
the sample is placed in an inductive coil which forms part of a tank circuit , which is self - resonant typically at @xmath15 and is driven by a tunnel diode circuit .
changes in the penetration depth or skin depth of the sample , caused by varying temperature , @xmath2 , applied dc magnetic field , @xmath16 , or angle between @xmath16 and the @xmath17 plane , @xmath18 , lead to changes in the coil inductance .
these are then detected as changes in resonant frequency , and converted to changes in @xmath19 via the relation @xmath20 $ ] , where @xmath21 is a geometrical factor set by the sample dimensions and @xmath22 refers to the resonant frequency without the sample .
the very high stability of the circuit , typically @xmath23 in @xmath15 , leads to a very high resolution of the order of a few .
the technique has been extensively validated through precise measurements in the cuprate superconductors of the non - linear meissner effect and of vortex parameters such as @xmath24 and pinning force constants@xcite .
in one experimental set - up the sample can be oriented with @xmath25 ( @xmath26 ) or @xmath27 ( @xmath28 ) , and fields up to @xmath29 can be applied . in another set - up
the angle @xmath18 can be varied continuously between @xmath30 and @xmath31 with an angular accuracy of @xmath32 . here
the maximum applied field is @xmath33 . in all cases the rf field @xmath34 and @xmath35 .
the single crystals of @xmath13 used for these measurements were grown using the ames lab @xmath36 flux growth method@xcite .
crystals grown via this method are in the form of plates with the crystallographic @xmath37 axis perpendicular to the surface of the plate .
the crystal used for these measurements has approximate dimensions @xmath38 . |
during the last decade intensive debates have centered around the question what happens when a system evolves from a band to a mott insulator @xcite in the context of different models in different dimensions @xcite . like the extended hubbard model @xcite and the ionic hubbard model @xcite in one dimension , or the ionic hubbard model @xcite in two dimensions .
a model that has gained a lot of attention in recent years is the bilayer hubbard model on a square lattice .
the discovery of bilayer band splitting in angle - resolved photoemission spectroscopy experiments @xcite for yba@xmath13cu@xmath14o@xmath15 ( ybco ) @xcite suggested the consideration of such a model as a minimal model for describing double - layered ybco compounds .
in fact , the bilayer hubbard model ( see fig . [ fig1:model ] ) was investigated by several groups within the dynamical mean field theory approximation ( dmft ) @xcite and cellular dmft @xcite .
transitions from metal to band insulator at small @xmath2 and from mott to band insulator at large @xmath2 were reported with increasing @xmath4 . however , by definition spatial fluctuations are completely ignored in dmft @xcite and such features like an intermediate non - fermi - liquid phase in the single - layered hubbard model at half filling @xcite are not captured by this approach . moreover ,
only a small cluster size of @xmath16 ( two sites in one layer and two sites in another layer ) was used in previous cellular dmft calculations for the bilayer hubbard model @xcite . for such sizes , the @xmath17 rotational symmetry of the square lattice
is broken , resulting in an artificially enhanced local pair within each plane as the interaction @xmath2 is switched on . as reported in ref .
, such a choice is unable to describe an intermediate orbital - selective phase in a two - orbital hubbard model . on the other hand , the dynamical cluster approximation ( dca ) with clusters sizes of @xmath18 ( ref . )
, the determinant quantum monte carlo method @xcite and the gutzwiller approximation @xcite -which do not suffer from the previous cluster limitations- were recently employed mainly to understand the nature of the superconducting state in the bilayer hubbard model away from half filling . in view of the above results , and due to the limitations of the various approaches used , there is still not a full understanding of the transition from band to mott insulating behavior as a function of @xmath4 in the bilayer hubbard model at half filling .
region studied here using dca is shown in darker colors .
, scaledwidth=40.0% ] in the present work we concentrate on this question and study the bilayer hubbard model at and away from half filling as the band splitting ratio @xmath4 is increased up to @xmath19 at a large interaction strength of @xmath20 in the framework of the dca @xcite .
we consider one plaquette in each layer , i.e. , a cluster size of @xmath21 , which keeps the rotation symmetry of the square lattice and allows for short - range spatial fluctuations .
we focus on the lower temperature regime , which could nt be accessed in previous studies @xcite due to the larger cluster sizes .
we use an interaction - expansion continuous - time quantum monte carlo algorithm as an impurity solver @xcite . note that the value of the critical interaction strength for the metal - insulator transition is sensitive to the level of approximation considered @xcite . within dca and for the cluster sizes considered in this work ,
the value @xmath20 is deep in the insulating phase . in order to distinguish between a mott and a band insulating regime
we analyze the behavior of ( i ) the density of states near the fermi level , ( ii ) the imaginary part of the self energy at the lowest matsubara frequency and ( iii ) the momentum resolved electron density . at small interlayer hopping @xmath4
we observe mott insulating behavior in the dca momentum sectors ( @xmath6)/(@xmath7 ) and band insulating behavior in the dca momentum sectors ( @xmath10)/(@xmath9 ) as was also obtained in cellular dmft with @xmath22 in the single - band hubbard model @xcite . with increasing @xmath4 the mott behavior
is rapidly suppressed in the dca momentum sectors ( @xmath6 ) and ( @xmath7 ) , and at @xmath8 it becomes dominant in the dca momentum sectors ( @xmath9 ) of the bonding band and ( @xmath10 ) of the anti - bonding band . at @xmath11 a tiny scattering rate
is observed in all momentum sectors at the fermi level , indicating a transition from a mott to a band insulator .
such a momentum - dependent evolution of the insulating state has not been reported in previous studies and will be analyzed in detail in the present work . and
( b ) 2.0 at half - filling in the weak coupling limit of @xmath23 .
the colored areas indicate the patches for momentum clusters @xmath24 , @xmath25 , @xmath26 , and @xmath27 of the dynamical cluster approximation with @xmath28.,scaledwidth=48.0% ] in particular , we find a monolayer plaquette singlet mott insulator ( m - psmi ) with strong intralayer plaquette order but weak interlayer antiferromagnetic ( af ) correlations at small @xmath4 and a bilayer plaquette singlet mott insulator ( b - psmi ) with strong af correlations between plaquettes belonging to different layers at intermediate values of @xmath4 . at the critical value @xmath11
we observe a transition from a mott to a band insulating state with tiny intralayer spin - spin correlations .
such a momentum - dependent phase behavior results from the competition and cooperation of short - range spatial fluctuations and interlayer coupling @xmath1 with the help of the coulomb interaction @xmath12 .
the paper is organized as follows . in sec .
[ formalism ] we present the model and dynamical cluster approximation . in sec .
[ results ] we present the density of states , the self - energy and electron density at each dca momentum sector as well as spin - spin correlations and we discuss the nature of mott insulator and band insulator at half - filling as well as non - fermi liquid behavior away from half- filling . finally , in sec .
[ conclusions ] we summarize our findings .
the bilayer hubbard hamiltonian can be written as @xmath29 where @xmath30 annihilates ( creates ) an electron with spin @xmath31 at site @xmath32 and layer @xmath33 , and @xmath34 is the chemical potential .
@xmath35 is the intralayer hopping matrix element between sites @xmath32 and @xmath36 in layer @xmath37 and @xmath1 is the interlayer hopping parameter which induces a band splitting into a bonding and an anti - bonding band . for @xmath38 ( @xmath37=1,2 ) with @xmath0 as energy unit throughout this paper , the energy dispersion is given as @xmath39 , where @xmath40 and a and b indices denote anti - bonding ( antisymmetric ) ( @xmath41 ) and bonding ( symmetric ) ( @xmath42 ) states , respectively @xcite .
the dca is the cluster extension of single - site dmft and the self - consistent equation can be written in momentum space with the assumption that the self - energy is constant in the brillouin zone sectors that are considered .
the cluster green s functions are calculated by integration of each sector : @xmath43 where @xmath44 is the number of @xmath45 points in each brillouin zone sector , @xmath34 the chemical potential , @xmath46 is the cluster momentum , @xmath47 is the dispersion relation for antibonding and bonding states , @xmath48 are the fermionic matsubara frequencies , and the summation over @xmath49 is performed in each brillouin zone sector . in our calculations , we considered a dca cluster with @xmath50 , where @xmath51 , @xmath26 , @xmath25 and @xmath27 ( see fig . [ fig1:sectors ] ) .
the converged self - energy @xmath52 is evaluated by means of eq .
( [ dcae ] ) and the dyson equation and we employed the interaction expansion continuous - time quantum monte carlo approach as an impurity solver @xcite .
all calculations presented below are for a temperature @xmath53 and more than @xmath54 qmc samplings are employed to measure the impurity green s function . in the non - interacting case ( @xmath23 ) , the model shows a band insulating state at @xmath55 due to the complete separation of bonding and anti - bonding bands , characterized by a formation of fully localized dimers between layers at half - filling .
in fig . [ fig1:dos ]
we show the density of states @xmath56 of the model at @xmath53 and @xmath20 for small , intermediate , and large band splittings with ratios @xmath57 , @xmath58 , and @xmath59 , respectively . at @xmath60
the gap is nonzero and dominated by the interaction @xmath12 , and the bonding and antibonding states both contribute to the lower and upper hubbard bands . on the other hand , at @xmath61 , the gap , which is also significant ,
is dominated by the band splitting @xmath62 . in this case
, only the bonding state contributes to the band below the fermi level while the antibonding one contributes to the band above the fermi level . at intermediate @xmath63
the gap amplitude is reduced to a smaller value compared to the gap amplitudes at @xmath57 and @xmath59 . here , both bonding and antibonding states significantly contribute to the bands above and below the fermi level .
however , the peak positions remain located at the same frequency as those for @xmath57 which indicates that mott behavior is still present , i.e. the peak position is only dependent on @xmath2 and not on @xmath64 . for chemical potentials @xmath65 with half - filling at @xmath20 and @xmath53 .
the band splitting is @xmath66 ( solid line ) , @xmath58 ( dashed line ) , and @xmath59 ( dotted line ) . the fermi level is at @xmath67.,scaledwidth=45.0% ] in order to obtain a deeper insight into the nature of the phases as a function of the band splitting , we analyze in the following the cluster self - energy at the various dca momentum sectors @xmath68 .
[ fig2:momentum ] ( a ) and ( b ) show the imaginary part of the dca cluster self - energy @xmath69 at the lowest matsubara frequency @xmath70 and the electron density @xmath71 , respectively , as a function of @xmath4 for @xmath20 at half filling where @xmath65 .
both quantities are shown for bonding ( b ) and antibonding ( a ) bands for the cluster momenta @xmath72 , @xmath73 ( identical to @xmath25 ) and @xmath27 at small band splitting @xmath74 , momentum selective mott insulating phases with large scattering rates @xmath75 are present in both bonding and antibonding bands at @xmath76 and @xmath26 momentum sectors where the electron density @xmath71 indicates almost half - filling , while the scattering rates at @xmath72 and @xmath27 momentum sectors remain small suggesting a band insulating behavior with empty or fully filled electron density @xmath77 in these sectors . further analysis on spin - spin correlations ( fig .
[ sup1 ] ) shows strong intralayer but weak interlayer antiferromagnetic correlations ( fig .
[ sup1 ] ( a ) and ( b ) ) .
moreover , the spin - spin correlation @xmath78 for different layers and nearest neighbor sites displays very weak ferromagnetic order at @xmath60 ( see fig .
[ sup1 ] ( c ) ) .
we denote the state at small @xmath64 m - psmi ( monolayer plaquette singlet mott insulator ) .
this m - psmi state has been reported to be present in the one - band hubbard model investigated by a 4-site cluster - dmft approach @xcite . at the lowest matsubara frequency @xmath70 and ( b ) electron density @xmath79 as a function of @xmath4 at each dca momentum sector @xmath80 .
results are shown at a temperature @xmath53 and an interaction strength @xmath20 for chemical potential @xmath81 ( half - filling @xmath82 ) .
@xmath46 are the dca cluster momenta and indices @xmath83 and @xmath84 stand for antibonding and bonding , respectively.,scaledwidth=45.0% ] as @xmath64 increases , the scattering rates @xmath75 at the @xmath76 and @xmath26 momentum sectors rapidly decrease towards zero while the scattering rate at the @xmath27 momentum sector of the bonding band ( see @xmath85 in fig .
[ fig2:momentum ] ( a ) ) and that at the @xmath86 momentum sector of the antibonding band ( see @xmath87 in fig .
[ fig2:momentum ] ( a ) ) develop dramatically . in terms of the electron density @xmath77 in fig .
[ fig2:momentum ] ( b ) , the @xmath27 momentum sector of the bonding band and the @xmath86 momentum sector of the antibonding band are filled with about one electron at @xmath88 .
the strong scattering is caused by interactions between electrons in the @xmath27 momentum sector of the bonding band and the @xmath86 momentum sector of the antibonding band at @xmath88 .
the spin - spin correlations exhibit intermediate intralayer as well as interlayer antiferromagnetic correlations .
specifically , the ferromagnetic correlations for different layers and nearest neighbor sites reach a maximum in fig .
[ sup1 ] ( c ) .
this means that the plaquette singlet orderings in each layer develop af correlations .
we denote this state bilayer plaquette singlet mott insulator ( b - psmi ) .
when @xmath4 is further increased , tiny scattering rates @xmath89 are observed in the @xmath27 momentum sector of the bonding band and the @xmath86 momentum sector of the antibonding band between @xmath90 and @xmath59 , while the electron densities @xmath71 are not fully occupied ( or empty ) as shown in figs .
[ fig2:momentum ] ( a ) and ( b ) .
such an insulator has been denominated a covalent band insulator @xcite . in the region @xmath91 ,
all scattering rates disappear and the system is in a band insulating state , where the electron density @xmath71 are fully filled ( or empty ) in all momentum sectors and all bands .
the ratio @xmath92 at which the mott to band insulator phase transition happens is smaller than the the ratio for the metal - insulator transition in the noninteracting case ( @xmath93 ) .
this is due to the fact that strong correlation narrows the bandwidth of the bonding and antibonding bands and consequently a smaller band splitting is required for opening a band gap .
the interlayer spin - spin correlations indicate a strong dimer state , while the remaining spin - spin correlations are extremely weak as shown in fig .
the state at @xmath93 is a band insulator with isolated dimers between layers . as a function of @xmath4 ( a ) between @xmath32 and nearest neighbors site @xmath94 in the same layer @xmath37 , ( b ) between layer @xmath37 and @xmath95 in the same site @xmath32 , and ( c ) between @xmath32 and nearest neighbor site @xmath94 as well as layer @xmath37 and @xmath95 for @xmath20 and @xmath53.,scaledwidth=45.0% ] the momentum - dependent evolution of the insulating states
can be well understood from the evolution of the fermi surface in the weak coupling limit as a function of @xmath1 .
as shown in fig . [ fig1:sectors ] ( a ) , at @xmath96 , all the fermi surfaces are located in the momentum sectors @xmath25 and @xmath26 , indicating that poles determined by @xmath97 are only present in these two sectors .
as the interaction @xmath12 becomes larger than the critical value of the mott metal - to - insulator transition , the opening of a gap at the fermi level indicates @xmath98 , which requires large scattering rates at the positions of the poles .
therefore , mott physics occurs in momentum sectors @xmath25 and @xmath26 . on the other hand , at @xmath99
( see fig .
[ fig1:sectors ] ( b ) ) , almost all the fermi surfaces from the bonding band enter the momentum sector @xmath27 while those from anti - bonding band are mostly in the momentum sector @xmath86 . in order to be a mott insulator at large interaction @xmath12 , large scattering rates
are again inevitable in the momentum sectors @xmath27 of the bonding band and @xmath86 of the anti - bonding band since poles are now located in these sectors .
thus , the momentum - dependent evolution of the insulating states is a consequence of the evolution of the fermi surface . at the lowest matsubara frequency @xmath70 as a function of @xmath4 .
results are shown at a temperature @xmath53 and an interaction strength @xmath20 for different values of the chemical potential : ( a ) @xmath100 ( heavily doped case ) and ( b ) 2.0 ( close to half - filled case ) .
the indices @xmath46 represent the dca cluster momenta and @xmath83 and @xmath84 indicate antibonding and bonding , respectively .
( c ) , ( d ) density of states @xmath56 for different chemical potentials @xmath101 at @xmath20 and @xmath53 .
the band splitting is ( c ) @xmath66 and ( d ) @xmath61 .
the fermi level is at @xmath67.,scaledwidth=45.0% ] we concentrate in what follows on the origin of non - fermi liquid or pseudogap , which has been discussed extensively in the literature @xcite , based on the bilayer hubbard model away from half - filling . in the heavily doped case ( @xmath102 ) ,
we observe at @xmath103 for @xmath53 a fermi liquid like metallic behavior with @xmath104 approaching small finite values due to the finite temperature effect ( see fig .
[ fig3:momentum ] ( a ) ) .
a quasiparticle peak is present at the fermi level in the dos ( see fig . [ fig3:momentum ] ( c ) ) .
close to the half - filled case , such as @xmath105 at @xmath103 , large but finite scattering rates are observed in the antibonding @xmath106 and the bonding @xmath106 sectors due to the enhancement of intralayer short - range af correlations . as a result
, a pseudogap appears in the dos at the fermi level ( see fig .
[ fig3:momentum ] ( c ) ) , reminiscent of the non - fermi liquid behavior observed in the single - band and multi - band hubbard models @xcite . at @xmath107
, though the scattering rates in all momentum sectors and all bands vanish due to the strong interlayer af correlations ( see fig .
[ fig3:momentum ] ( b ) ) , pseudogaps exist in both heavily doped and nearly half - filled cases , indicating that a strong scattering rate is not a necessary condition for the appearance of non - fermi liquid behavior ; rather the short - range af correlations alone can be responsible for the non - fermi liquid behavior .
furthermore , in order to confirm the transitions from the fermi liquid to mott insulator via non - fermi liquid as a function of electron doping ( or chemical potential ) , we plot the @xmath108 at the lower temperature @xmath109 and @xmath103 for @xmath20 in fig .
[ fig6:selfenergy ] . in the case of @xmath110
, @xmath111 converges to almost zero indicating a fermi - liquid behavior . as the electron density
@xmath112 approaches half - filling between @xmath113 and @xmath114 , @xmath111 converges to finite values .
such a behavior is typical for non - fermi liquid states .
finally , at half - filling with @xmath115 the system is a mott insulator with @xmath116 exhibiting a diverging behavior . as a function of matsubara frequency @xmath48 for @xmath109 , @xmath66 , and @xmath20 in different electron densities @xmath112.,scaledwidth=45.0% ]
in conclusion , we investigated a two - dimensional bilayer hubbard model on the square lattice as a function of band splitting and doping at a large interaction value @xmath20 by means of the dynamical cluster approximation with a @xmath21 site cluster with short - range spatial as well as quantum fluctuations .
the scattering rate , which indicates the degree of mott physics , rapidly disappears in the momentum sectors ( @xmath6 ) and ( @xmath117 ) with increasing interlayer hopping @xmath4 .
in fact we find a momentum - selective phase reentrant behavior from band insulating states at weak interlayer hopping @xmath118 to mott insulating behavior at @xmath119 and then from mott insulating to band insulating behavior at strong interlayer hopping @xmath120 .
these transitions are identified from the scattering rates in the ( @xmath9 ) momentum sector of the bonding band and the ( @xmath10 ) momentum sector of the anti - bonding band at half filling .
interesting phases are established with two consecutive phase transitions from a monolayer plaquette singlet mott insulator ( m - psmi ) to a band insulator through an intermediate phase , called bilayer plaquette singlet mott insulator ( b - psmi ) , where mott physics is more present in the @xmath27 sector of the bonding band and the @xmath86 of the anti - bonding band , rather than in the @xmath25 and @xmath121 sectors as usually observed in an antiferromagnetic mott insulator .
we attribute the unusual consecutive phases to competition and cooperation between short - range spatial correlations with quantum fluctuations and interlayer hopping @xmath4 .
the transition of mott to band insulator with the absence of large scattering rates in all momentum sectors is found at large interlayer hopping @xmath11 .
furthermore , since the momentum - dependent evolution of the insulating behavior with @xmath4 is strongly controlled by the evolution of the fermi surface in the weak - coupling limit , we expect that consideration of larger clusters and different geometries in dca wo nt change this scenario qualitatively .
finally we also find that away from half filling , non - fermi liquid behavior is dominated by antiferromagnetic correlations rather than the finite scattering rate at the fermi level .
we suggest that this non - fermi liquid behavior might be related to anomalous phenomena like the fermi arc or hole pocket .
we expect that momentum - selective phenomena may exist in many cases which calls for further studies in various models and real materials where the short - range spatial fluctuations are emphasized .
we would like to thank j. kune , l. f. tocchio , f. becca , i.i .
mazin , m. capone , and c. gros for useful discussion .
we gratefully acknowledge financial support from the deutsche forschungsgemeinschaft through grants for 1346 and sfb / trr 49 .
y.z is supported by national natural science foundation of china ( no . 11174219 ) , shanghai pujiang program ( no .
11pj1409900 ) , research fund for the doctoral program of higher education of china ( no .
20110072110044 ) and the program for professor of special appointment ( eastern scholar ) at shanghai institutions of higher learning .
h.l is supported by erc / fp7 through the starting independent grant `` superbad '' , grant agreement no .
240524 .
99 n. f. mott , proceedings of the physical society of london series a * 62 * , 416 ( 1949 ) .
f. gebhard , _ the mott metal - insulator transition _ springer tracts in modern physics * 137 * springer ( 1997 ) .
m. nakamura , phys .
b * 61 * , 16 377 ( 2000 )
. y. z. zhang , phys .
lett . * 92 * , 246404 ( 2004 ) .
tam , s .- w .
tsai , d. k. campbell , phys .
lett . * 96 * , 036408 ( 2006 ) .
s. ejima , s. nishimoto , phys .
* 99 * , 216403 ( 2007 )
. m. fabrizio , a. o. gogolin , a. a. nersesyan , phys .
lett . * 83 * , 2014 ( 1999 ) .
y. z. zhang , c. q. wu , h. q. lin , phys . rev .
b * 67 * , 205109 ( 2003 ) . c. d. batista , a. a. aligia , phys . rev . lett . * 92 * , 246405 ( 2004 ) .
ara go , gun sang jeon , phys .
b * 84 * , 195102 ( 2011 ) .
a. garg , h. r. krishnamurthy , m. randeria , phys .
lett . * 97 * , 046403 ( 2006 ) .
s. s. kancharla , e. dagotto , phys .
lett . * 98 * , 016402 ( 2007 ) .
n. paris , k. bouadim , f. hebert , g. g. batrouni , r. t. scalettar , phys .
lett . * 98 * , 046403 ( 2007 ) .
chen , h. zhao , h .- q .
lin , c .- q .
w , new journal of physics * 12 * , 093021 ( 2010 ) .
l. craco , p. lombardo , r. hayn , g. i. japaridze , and e. muller - hartmann , phys .
b * 78 * , 075121 ( 2008 ) .
a. damascelli , z. hussain , z - x .
shen , rev .
75 * , 473 ( 2003 ) .
d. fournier et al .
nature physics * 6 * , 905 ( 2010 ) .
a. fuhrmann , d. heilmann , h. monien , phys .
b * 73 * , 245118 ( 2006 ) .
h. hafermann , m. i. katsnelson , a. i. lichtenstein , euro .
85 * , 37006 ( 2009 ) .
kancharla , s. okamoto , phys .
b * 75 * , 193103 ( 2007 ) .
a. georges , g. kotliar , w. krauth , m. j. rozenberg , rev .
phys . * 68 * , 13 ( 1996 ) .
y. z. zhang , m. imada , phys .
b * 76 * , 045108 ( 2007 ) .
h. park , k. haule , g. kotliar , phys .
* 101 * , 186403 ( 2008 ) .
e. gull , p. werner , w. wang , m. troyer , a. millis , europhys .
* 84 * , 37009 ( 2008 ) .
h. lee , y.z .
zhang , h.o .
jeschke , r. valenti , h. monien , phys .
lett . * 104 * , 026402 ( 2010 ) .
maier , d.j .
scalapino , phys .
b * 84 * , 180513(r ) ( 2011 ) .
k. bouadim , g.g .
batrouni , f. hebert , r.t .
scalettar , phys .
b * 77 * , 144527 ( 2008 ) .
n. lanata , p. barone , m. fabrizio , phys .
b * 80 * , 224524 ( 2009 ) .
m. h. hettler , a. n. tahvildar - zadeh , m. jarrell , t. pruschke , and h. r. krishnamurthy , phys .
b * 58 * , 7475(r ) ( 1998 ) . t. maier , m. jarrell , t. pruschke , m.h .
hettler , rev .
* 77 * , 1027 ( 2005 ) a.n .
rubtsov , v.v .
savkin , a.i .
lichtenstein , phys .
b * 72 * , 035122 ( 2005 ) .
f.f . assaad , t. c. lang , phys .
b * 76 * , 035116 ( 2007 ) .
e. gull , a. j. millis , a. lichtenstein , a. rubtsov , m. troyer , p. werner , rev .
phys . * 83 * , 349 ( 2011 ) .
michael sentef , jan kune , philipp werner , arno p. kampf , phys .
b * 80 * , 155116 ( 2009 ) . l. f. tocchio , h. lee , h. o. jeschke , r. valenti , c. gros phys . rev .
b * 87 * , 045111 ( 2013 ) .
t. d. stanescu , g. kotliar , phys .
b * 74 * , 125110 ( 2006 )
. s. sakai , y. motome , m. imada , phys .
* 102 * , 056404 ( 2009 ) .
e. gull , m. ferrero , o. parcollet , a. georges , a. j. millis , phys .
b * 82 * , 155101 ( 2010 ) .
g. sordi , k. haule , a .- m.s .
tremblay , phys .
b * 84 * , 075161 ( 2011 ) .
s. sakai , g. sangiovanni , m. civelli , y. motome , k. held , m. imada , phys . rev .
b * 85 * , 035102 ( 2012 ) .
h. lee , y .- z .
zhang , h.o .
jeschke , r. valenti , phys .
b * 84 * , 020401(r ) ( 2011 ) . | we investigate the nature of the insulating phases in a bilayer hubbard model with intralayer coupling @xmath0 and interlayer coupling @xmath1 at large interaction strength @xmath2 and half - filling .
we consider a dynamical cluster approximation with a cluster size of @xmath3 , where short - range spatial fluctuations as well as on - site dynamical fluctuations are emphasized . by varying the band splitting ( @xmath4 )
, we find that at @xmath5 the mott behavior is rapidly suppressed in the momentum sectors ( @xmath6 ) and ( @xmath7 ) . at @xmath8 mott features dominate in the momentum sectors ( @xmath9 ) of the bonding band and ( @xmath10 ) of the anti - bonding band and at @xmath11 a tiny scattering rate is observed in all momentum sectors at the fermi level , indicating a transition from a mott to a band insulator .
we attribute such a momentum - dependent evolution of the insulating behavior to the competition and cooperation between short - range spatial fluctuations and interlayer coupling @xmath1 with the help of the coulomb interaction @xmath12 .
finally , we also discuss the possible appearance of non - fermi liquid behavior away from half - filling . |
thanks to the efforts of babar and belle collaborations in the past years , most of the @xmath9 mesons decays with branching fractions @xmath10 have been measured . with the particle physics entering the lhc era ,
more rare decays of @xmath11 mesons , especially of @xmath12 meson , are expected to be well measured .
in addition , most recently , the upgrading superkekb / belle - ii experiment has started test operations and succeeded in circulating and storing beams in the electron and positron rings .
so , in the near future , the measurements of @xmath11 meson decays are expected to reach unprecedented precision , which will provide a much more fertile ground for testing the flavor picture of the standard model ( sm ) and exploring underlying mechanisms . for the nonleptonic two - body @xmath11 meson decays , the theoretical evaluation is generally complicated due to the nontrivial qcd dynamics related to the hadronic final states . in order to evaluate the strong interaction corrections to the amplitude , several attractive qcd - inspired approaches , including qcd factorization ( qcdf ) @xcite , the pqcd approach @xcite and the soft - collinear effective theory ( scet ) @xcite , have been presented . for the case of two light final states ,
the theoretical evaluations with qcdf approach have been fully developed in the heavy quark limit , for instance refs .
for the case of heavy - light final states , such as @xmath13 and @xmath14 _ et al .
_ , the calculation is generally much more complicated due to the un - negligible @xmath15 quark mass . in refs .
@xcite , the factorization formula at two - loop order has been proven , and the explicit results of qcd corrections for @xmath16 , @xmath17 , @xmath18 and longitudinally polarized @xmath19 ( @xmath20 and @xmath21 are light pseudoscalar and vector mesons , respectively ) decays have been presented .
the @xmath22 induced nonleptonic @xmath11 decays , which are tree - dominated and ckm - favored , have relatively large branching fractions and have been widely studied in various theoretical frameworks , for instance refs .
@xcite . in the previous works for the @xmath23 decays based on the qcdf , the qcd corrections to the longitudinal amplitude have been fully evaluated at next - to - leading order ( nlo ) @xcite and next - to - next - to - leading order ( nnlo ) @xcite . even though the @xmath23 decays are dominated by the longitudinally polarized final states , to keep fairness and consistence , the power - suppressed transverse amplitudes
should also be calculated to the same order as the longitudinal one , which is also essential for relatively accurate theoretical results , especially for the polarization fractions .
that is what we would like to do in this paper .
moreover , due to the refined experimental measurements , it is also worth performing detailed phenomenological analyses and testing whether the data and the theoretical results are in agreement .
in addition to @xmath11 mesons , their excited states , such as @xmath24 mesons with quantum number of @xmath25 and @xmath26 , also could decay through the same transitions as @xmath11 mesons at quark level .
thanks to the rapid development of heavy - flavor experiments @xcite , even though @xmath24 decays are dominated by the electromagnetic processes @xmath27 , the @xmath24 weak decays with branching fractions @xmath28 are still hopeful to be observed by belle - ii as analyzed in refs .
moreover , owing to the much larger beauty production cross section of @xmath29 collisions @xcite , the lhc experiments may also provide a lot of experimental information for @xmath24 decays , such as the leptonic @xmath30 decay with branching fraction @xmath31 analyzed in ref .
recently , a few interesting theoretical studies of @xmath24 weak decays have been made , for instance refs .
@xcite . in our previous work
@xcite , the @xmath32 decays have been studied in the framework of naive factorization ( nf ) . in this paper , the qcd corrections at nlo
will be evaluated with qcdf approach for relatively accurate prediction , and the phenomenological studies will be updated simultaneously .
our paper is organized as follows . in section 2 , the theoretical framework and calculations for @xmath33 and @xmath34 decays
are presented with qcdf approach .
section 3 is devoted to the numerical results and discussions .
finally , we give our summary in section 4 .
the effective hamiltonian responsible for the @xmath35 ( @xmath36 ) induced @xmath33 and @xmath37 decays could be written as @xmath38 where @xmath39 is the fermi coupling constant , @xmath40 is the product of ckm matrix elements , and @xmath41 are local tree four - quark operators defined as @xmath42 the corresponding wilson coefficients @xmath43 summarize the physical contributions above scale of @xmath44 and are calculable with the perturbation theory @xcite . in order to obtain the decay amplitudes
, the remaining work is to accurately calculate the hadronic matrix elements of every local operators in effective hamiltonian .
the simplest way to evaluate the hadronic matrix elements is the nf scheme @xcite .
however , in the framework of nf , the amplitudes are renormalization - scale - dependent , and the non - factorizable contributions dominated by the hard gluon exchange are lost . in order to remedy these deficiencies ,
the qcdf approach is proposed by bbns @xcite . in the framework of qcdf , up to power corrections of order @xmath45 , the hadronic matrix elements @xmath46 ( @xmath47 is heavy and @xmath48 is light ) obey the factorization formula @xcite , @xmath49 where @xmath50 is a @xmath51 form factor ; @xmath52 is the light - cone distribution amplitude ( lcda ) for the quark - antiquark fock state of meson @xmath48 ; and @xmath53 denotes the hard - scattering function , which is calculable order by order from the first principle of perturbative qcd theory . applying the qcdf formula ,
the amplitude of @xmath54 decay could be written as @xmath55 where @xmath56 denotes the helicity of @xmath21 meson ; @xmath57 is the product of matrix elements of current operators , _
i.e. _ , @xmath58 ; @xmath59 is the effective flavor coefficient and includes the nonfactorizable contributions . without the qcd corrections , the nf result , @xmath60 , is recovered .
the amplitude of @xmath61 decay is obtained from the formula above by replacing @xmath62 and @xmath63 .
the explicit expressions of @xmath57 and @xmath59 are given in the following .
the decay constant of emitted vector meson is defined through the current matrix element , @xmath64 where @xmath65 and @xmath66 denote the mass and the polarization vector , respectively . meanwhile , with the same conventions as ref .
@xcite , the form factors are defined by @xmath67,\end{aligned}\ ] ] for @xmath68 transition , and @xmath69,\end{aligned}\ ] ] for @xmath70 transition , where @xmath71 , @xmath72 is the polarization vector of @xmath73 meson , and the sign convention @xmath74 is taken .
then , after contracting the current matrix elements , we finally obtain @xmath75\,,\\ \label{eq : hmpb } h_{\mp}&=&i\,f_v\,m_v\left[(m_b+m_{d^*})a_1(m_v^2)\pm \frac{2\,m_b\,p_c}{m_b+m_{d^*}}v^{b\to d^*}(m_v^2)\right]\end{aligned}\ ] ] for @xmath54 decays , and @xmath76\,,\\ \label{eq : hmpbstar } h_{\mp}^{\prime}&=&-i\,f_v\,m_v\left[(m_{b^*}+m_{d})a_1^{b^*\to d}(m_v^2)\pm \frac{2\,m_{b^*}\,p_c^{\prime}}{m_{b^*}+m_{d}}v^{b^*\to d}(m_v^2)\right]\end{aligned}\ ] ] for @xmath61 decays , in which , @xmath77[m_b^2-(m_{d^*}-m_v)^2]}}{2m_b}\end{aligned}\ ] ] and @xmath78 is obtained from @xmath79 by replacing @xmath80 and @xmath81 .
the effective coefficient @xmath59 in the amplitude , eq . ( [ amp ] ) , includes the nonfactorizable contributions from qcd radiative vertex corrections ( the penguin diagrams do not contribute to @xmath82 and @xmath83 decays at the order of @xmath84 ) , and could be written as @xmath85 after calculation , we get the explicit expressions of the vertex corrections @xmath86 written as @xmath87\,,\\ \label{eq : vmp } v_1^{-,+ } = { \int}_{0}^{1}du\ , \,{\phi}_{b , a}(u ) \left [ 3\,{\log } \big ( \frac { m_{b}^{2 } } { { \mu}^{2 } } \big)+ 3\,{\log } \big ( \frac { m_{c}^{2 } } { { \mu}^{2 } } \big)- 18 + g_{-,+}(u)\right]\ , , \end{aligned}\ ] ] where @xmath88 is the leading - twist lcda and conventionally expanded in gegenbauer polynomials @xcite , @xmath89\,;\end{aligned}\ ] ] @xmath90 are the twist-3 lcdas given by @xmath91 it could be found that only the leading - twist lcda of emitted vector meson contributes to @xmath92 and twist-3 ones contribute to @xmath93 .
in addition , the loop functions @xmath94 in eqs .
( [ eq : v10 ] ) and ( [ eq : vmp ] ) are written as @xmath95-\zeta(r_c)\,,\\ \label{eq : gt } g_{\mp}(u)&= & \frac{1+c_{a}}{1-c_{a}}\,{\log}(c_{a})-\frac{4\,c_{b}}{1-c_{b}}\,{\log}(c_{b } ) + \frac{1+c_{d}}{1-c_{d}}\,{\log}(c_{d})-\frac{4\,c_{c}}{1-c_{c}}\,{\log}(c_{c } ) \nonumber \\ & & + f(c_{a})-f(c_{b})-f(c_{c})+f(c_{d } ) + 2\ , { \log}(r_{c}^{2 } ) \big [ { \log}(c_{a } ) -{\log}(c_{b } ) \big]-\xi_{\mp}(r_c)\ , , \end{aligned}\ ] ] in which , @xmath96 , @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 in the eqs .
( [ eq : g0 ] ) and ( [ eq : gt ] ) , the functions @xmath102 and @xmath103 contain all of the anti - symmetrical contributions under the transformation @xmath104 ( or @xmath105 ) .
they are written as @xmath106 -r_{c}^{-1}\,\big [ \frac{c_{d}}{(1-c_{d})^{2}}\,{\log}(c_{d})+ \frac{1}{1-c_{d } } \big]\ , , \\
\label{xi } \xi_{\mp}(r_c)&=&k_{\mp}\ , r_{c}\ , \big [ \frac{2c_{a}-1}{(1-c_{a})^{2}}\,{\log}(c_{a})+ \frac{1}{1-c_{a } } \big ] + k_{\mp}\,r_{c}^{-1}\,\big [ \frac{2c_{d}-1}{(1-c_{d})^{2}}\,{\log}(c_{d})+ \frac{1}{1-c_{d}}\big]\ , , \end{aligned}\ ] ] in which , @xmath107 with @xmath108 . in the limit of @xmath109 , both @xmath102 and @xmath103 vanish , i.e. , @xmath110 .
moreover , after taking the limit @xmath109 , one can find that the results of @xmath111 decays , which have been presented in the eqs .
( a.7 ) and ( a.8 ) of ref .
@xcite , could be recovered from eqs .
( [ eq : v10 ] ) , ( [ eq : vmp ] ) , ( [ eq : g0 ] ) and ( [ eq : gt ] ) . for @xmath112 , the only difference between longitudinally polarized @xmath113 and @xmath114 ( @xmath115 is a light meson ) decays is the overall sign of @xmath102 ( or the sign of @xmath116 ) , which has been pointed out in ref .
@xcite and confirmed in ref .
. such difference could be easily understood from that : ( i ) after computing the one - loop correction , the ( @xmath117 ) pair ( @xmath118 are light quarks ) always retains its @xmath119 structure , but the ( @xmath120 ) pair has not only @xmath119 but also @xmath121 structure due to the un - negligible @xmath122 ( in this paper , the contributions of the later are exactly collected into the functions @xmath102 and @xmath103 , _ i.e. _ , eqs .
( [ zeta ] ) and ( [ xi ] ) ) ; ( ii ) only the @xmath21-current contributes to @xmath114 and only the @xmath123-current contributes to longitudinally polarized @xmath113 .
one can also refer to the section 4.4 in ref .
@xcite for detailed explanation .
our result of @xmath92 is in consistence with the result in ref .
@xcite , while the transverse results @xmath124 and @xmath125 are first presented . for @xmath125 , because both @xmath21- and @xmath123-currents contribute to the transversely polarized @xmath126 decay , the overall factor @xmath127 instead of overall sign exists in eq .
( [ xi ] ) .
in addition , it should be noted that the strong phase can be obtained by recalling that @xmath128 is @xmath129 with @xmath130 infinitesimal . with the amplitudes given above , the branching fraction of @xmath131 decay
is defined as @xmath132 where @xmath133 is the total decay width of @xmath134 meson . for @xmath135 decays
, the definition is obtained from eq .
( [ br1 ] ) by replacing @xmath62 , @xmath136 , @xmath137 and multiplying by an additional factor @xmath138 , which is caused by averaging over the spin of initial state @xmath139 .
besides of the branching fraction , the polarization fractions are also very important observables , which are defined as @xmath140 where @xmath141 and @xmath142 are parallel and perpendicular amplitudes , and could be easily gotten through @xmath143 .
before presenting our numerical results , we would like to clarify the input parameters used in the evaluations . for the ckm matrix elements ,
we adopt the wolfenstein parameterization @xcite and choose the parameters @xmath123 and @xmath144 as @xcite @xmath145 for the well - known fermi coupling constant @xmath146 , the masses of mesons and the total decay widths ( or lifetimes ) of @xmath11 mesons , we take the central values given by pdg @xcite . however , for @xmath147 , there is no available experimental and theoretical information at present . because the electromagnetic processes @xmath148 dominate the decays of @xmath24 mesons , we take the approximation @xmath149 in our evaluations .
the theoretical predictions for @xmath150 have been given in various theoretical models @xcite .
in this paper , we take the central values of the latest results @xcite @xmath151 which are in agreement with most of the other theoretical predictions . as for the light mesons decay constants and gegenbauer moments ( at @xmath152 ) , we take @xcite @xmath153 then , the residual inputs are the qcd form factors @xmath154 and @xmath155 , which are crucial for evaluating the observables of nonleptonic @xmath156 decays . for the @xmath157 transition , the form factors ( or the relevant parameters ) could be precisely extracted from the well - measured @xmath158 decay distributions . after performing a four - dimensional fit to the measurements of exclusive @xmath158 decays
, the hfag presents the averaged results of the caprini , lellouch and neubert ( cln ) @xcite form factor parameters for @xmath157 transition @xcite @xmath159 the qcd form factors @xmath154 and @xmath155 are obtained through the relation @xcite @xmath160 where the ratio @xmath161 and the kinematical variable @xmath162 .
the @xmath163 dependence of @xmath164 and @xmath165 reads @xcite @xmath166\,,\\ r_1(\omega)&=&r_1(1)-0.12(\omega-1)+0.05(\omega-1)^2\,,\\ r_2(\omega)&=&r_2(1)+0.11(\omega-1)-0.06(\omega-1)^2\,,\end{aligned}\ ] ] with @xmath167 .
for the form factors of @xmath168 and @xmath169 transitions , due to the lack of the experimental information , we employ the bauer - stech - wirbel ( bsw ) model @xcite for estimating their values . with the constituent masses
@xmath170 , @xmath171 , @xmath172 , @xmath173 and @xmath174 , we get @xmath175 to be conservative , @xmath176 uncertainties are assigned to the values above . with the assumption of the nearest pole dominance , the @xmath177 dependences of form factors read @xcite @xmath178 where @xmath179 is the state of @xmath180 with quantum number of @xmath181 ( @xmath182 and @xmath20 are the quantum numbers of total angular momenta and parity , respectively ) . [
fig : sd ] .
the nlo results of effective coefficient @xmath183 at three different renormalization scales . the lo ( nf ) results are also listed for comparison . [ cols="<,^,^,^,^",options="header " , ] [ tab : bstar ] in table [ tab : b ] , we have summarized our numerical results for the observables of @xmath131 decays .
moreover , the experimental data and the results of some previous works are also listed in table [ tab : b ] for comparison . in table
[ tab : bstar ] , our theoretical predictions for the observables of @xmath184 decays are presented .
the theoretical uncertainties of our results in these tables are obtained through separately evaluating the uncertainty induced by each input parameter and then adding the individual uncertainties in quadrature .
the followings are some analyses and discussions . using the approximation of @xmath185 and @xmath186 , eqs .
( [ eq : h0b ] ) and ( [ eq : hmpb ] ) could be reduced to @xmath187\,,\\ h_{\mp}&\approx&i\,f_v\,m_v\left[(m_b+m_{d^*})a_1^{b\to d^*}\pm(m_b - m_{d^*})v^{b\to d^*}\right]\,.\end{aligned}\ ] ] similarly , for @xmath188 decays , eqs .
( [ eq : h0bstar ] ) and ( [ eq : hmpbstar ] ) could be simplified as @xmath189\,,\\ h_{\mp}^{\prime}&\approx&-i\,f_v\,m_v\left[(m_{b^*}+m_{d})a_1^{b^*\to d}\pm ( m_{b^{*}}-m_{d})v^{b^*\to d}\right]\,.\end{aligned}\ ] ] from the simplified expression given above , one can obtain the relation @xmath190 for @xmath54 decays and @xmath191 for @xmath61 decays , which implies the dominance of the longitudinal polarization in both @xmath54 and @xmath61 decays .
our numerical results for the polarization fractions listed in tables [ tab : b ] and [ tab : bstar ] , @xmath192 $ ] , fulfill such expectation . for the measured @xmath193 and @xmath194 decays ,
our results of the polarization fractions are in a good agreement with the data and the theoretical results in previous works . for the branching fractions
, it could be found from table [ tab : b ] that our results are in consistence with the ones based on the instantaneous bether - salpeter method ( bsm ) @xcite , the heavy quark symmetry ( hqs ) @xcite and the qcdf with the nnlo corrections to the longitudinal polarization amplitude @xcite , as well as the other theoretical results in refs .
@xcite , but a bit larger than the result of the pqcd approach @xcite .
in addition , the most recent updated pqcd results , @xmath195 and @xmath196 @xcite ( the results of @xmath197 decays are not updated ) , agree well with ours . compared with the experimental data ,
it could be found from table [ tab : b ] that : ( i ) our qcdf result for @xmath198 is in a good agreement with the data ; ( ii ) however , the results ( central values ) for @xmath199 and @xmath200 are much larger than the data .
moreover , such deviation would be further enlarged if the nnlo qcd corrections , which provide about @xmath201 enhancement to the branching fractions , are included @xcite .
in addition , the results within the other theoretical frameworks also deviate from the data more or less , which can be seen exactly from table [ tab : b ] . to clarify such possible mismatch
, one can define the quantity @xmath202 comparing with the @xmath203 ( @xmath2 ) decay appeared in denominator , the @xmath204 decay appeared in numerator receives additional weak - annihilation corrections , which is however power - suppressed and numerically trivial for the tree - dominated decays @xcite .
so , in the limit of u - spin flavor symmetry , the result @xmath205 is expected . using the experimental data listed in table [ tab : b ] and the error transfer formula , we get @xmath206=0.70\pm0.18~(0.57^{+0.17}_{-0.18})\,,\ ] ] in which the number in the round brackets is the result gotten by using the direct measurement @xmath207 @xcite instead of pdg result @xmath208 .
the results in eq .
( [ eq : rdsexp ] ) are much smaller than 1 , and imply the significant effect of u - spin flavor - symmetry - breaking .
however , the current theoretical results , @xmath209\simeq1.09\,{\rm(this~work)}\,,1.44\,{\rm(pqcd)}\,,1.14{\rm(bsm)}\,,0.98\,{\rm(hqs)}\,,1.44\,{\rm(nnlo)}\end{aligned}\ ] ] are much larger than the experimental data , eq .
( [ eq : rdsexp ] ) , at about @xmath210 , @xmath211 , @xmath212 , @xmath213 and @xmath211 level , respectively .
if the future refined measurements , especially on @xmath194 decay , confirm the large u - spin flavor - symmetry - breaking effects illustrated by eq .
( [ eq : rdsexp ] ) , it would be a serious challenge to the current theoretical estimation . besides @xmath214
, the measurements on @xmath215 also could provide a judgment for the possible unexpected large flavor - symmetry - breaking .
numerically , the theoretical results are @xmath216\simeq1.07\,{\rm(this~work)}\,,1.44\,{\rm(pqcd)}\,,1.14{\rm(bsm)}\,,0.94\,{\rm(hqs)}\,,1.42{\rm(nnlo)},\end{aligned}\ ] ] which are very close to the results for @xmath214 given by eq .
( [ eq : rdsrhoth ] ) , _ i.e. _ , @xmath217\simeq r_{ds}^{\rho}[{\rm theo.}]$ ] .
in addition to above - mentioned `` @xmath214 puzzle '' , one may find another tension between the theoretical results and the data .
firstly , we would like to emphasize that the values of cln form factor parameters for @xmath157 transition are purely extracted from the experimental measurements of @xmath218 decays , and therefore , are model - independent . with such inputs ,
we obtain @xmath219 and @xmath220 , where the theoretical uncertainties are well controlled due to the precise measurements of @xmath218 decays .
however , unfortunately , such theoretical results are about @xmath221 and @xmath222 , respectively , larger than the data . due to the fact that the new physics ( np ) corrections are generally trivial for the branching fractions of ckm - favored and tree - dominated @xmath223 decays , the large gap between the theoretical estimation and experimental data for @xmath224 , which is called `` @xmath4 puzzle '' in the following discussions for convenience , is hardly to be moderated by the np contribution . in order to check
if the `` @xmath4 puzzle '' is stable , an improved way is to perform an measurement on the ratio defined by @xmath225 in which @xmath226 .
firstly , we would like to estimate its theoretical result . neglecting the lepton mass ,
the differential decay rate can be written as @xmath227 where @xmath228 are the helicity amplitudes and are @xmath177-dependent .
one may refer to refs .
@xcite _ et al .
_ for the details . at @xmath229
, we get @xmath230 where @xmath231 have been given by eqs .
( [ eq : h0b ] ) and ( [ eq : hmpb ] ) , and the explicit expression for @xmath232 could be found in ref .
then , further considering that @xmath233 numerically , we finally obtain @xmath234 which is independent of the form factors , and principally could be precisely determined .
numerically , for @xmath193 and @xmath235 decays , one can easily get the theoretical prediction @xmath236}&\simeq & 2.91\,{\rm gev^2}\cdot\left[\frac { f_\rho}{0.216\,{\rm gev}}\right]^2\left[\frac{|v_{ud}|}{0.9743}\right]^2\left[\frac{|\alpha_1|}{|1.053 + 0.010i|}\right]^2\,,\\ \label{eq : rvlth2 } r_{k^*/\ell\bar{\nu}_\ell}{\rm [ theo.]}&\simeq & 0.16\,{\rm gev^2}\cdot\left[\frac { f_{k^*}}{0.220\,{\rm gev}}\right]^2\left[\frac{|v_{us}|}{0.2254}\right]^2\left[\frac{|\alpha_1|}{|1.053 + 0.010i|}\right]^2\ , .
\end{aligned}\ ] ] using the most recent result @xmath237 at nnlo in qcdf @xcite , one can obtain the similar results @xmath238 and @xmath239 .
then , using the distribution of @xmath240 measured by belle collaboration @xcite and the data of @xmath199 and @xmath200 listed in table [ tab : b ] , we obtain the experimental results @xmath241}&=&2.13\pm0.34\,,\\ \label{eq : rvlexp2 } r_{k^*/\ell\bar{\nu}_\ell}{\rm[exp.]}&=&0.10\pm0.02\ , . \end{aligned}\ ] ] in the estimation , to get the differential decay rate at @xmath229 , we pick out the measurement at the bin @xmath242 $ ] , which covers the point @xmath243 , and take the approximation @xmath244}}{1.50 - 1.45}$ ] . comparing eqs .
( [ eq : rvlth1 ] ) and ( [ eq : rvlth2 ] ) with eqs . ( [ eq : rvlexp1 ] ) and ( [ eq : rvlexp2 ] ) , one may find that the sm expectations for @xmath245 and @xmath246 deviate from the experimental results by about @xmath247 and @xmath248 , respectively . in ref .
@xcite , the authors have also pointed out that the the deviation is at the level of @xmath249 to the nnlo accuracy .
a specific experimental measurement or analysis on @xmath5 at a very narrow bin covering @xmath243 is required for a much more reliable result . for @xmath250 decays , it could be found from table [ tab : bstar ] that all decays have the branching fractions @xmath28 , and therefore , are hopeful to be measured by belle - ii experiment , which have been pointed out in ref .
moreover , they are also in the scope of lhcb experiment , which can be seen from the following analysis . to make an estimate , we take @xmath251 decay as a reference . using
the data corresponding to integrated luminosities of @xmath252 of @xmath29 collisions at a center - of - mass energy of @xmath253 , about @xmath254 @xmath251 decay events have been measured by lhcb collaboration @xcite .
after high - luminosity upgrade , a data sample of @xmath255 will be collected by lhcb collaboration at a much higher @xmath256 , which results in a further enhancement of @xmath257 production by a factor about 2 @xcite . moreover
, the most of @xmath11 mesons detected at lhc are mainly produced through @xmath148 decays because @xmath24 mesons are often produced by about 3 times more than the @xmath11 mesons @xcite .
finally , further considering @xmath258 , we estimate that of the order of @xmath259 @xmath260 decay events could be collected by lhcb .
furthermore , taking @xmath261 as reference instead of @xmath251 and revisiting the estimate above , one may find about @xmath262 @xmath263 decay events are expected to be observed in the high - luminosity lhc era .
the @xmath264 decays occur with the same transition as @xmath265 decays at quark level , and have similar amplitudes as @xmath265 decays .
so , once measured in the future , they are expected to provide crosschecking of the possible @xmath3 and @xmath4 ( or @xmath5 ) puzzles mentioned above .
for this point , one can define the ratios @xmath266 which are similar to the ones for @xmath11 decays , eqs .
( [ eq : rds ] ) and ( [ eq : rvl ] ) . for @xmath267
, it is expected to be equal to @xmath5 , eq .
( [ eq : rvlsim ] ) . for @xmath7 ,
numerically , we get @xmath268=\frac{\tau_{b_s^*}}{\tau_{b^*}}\cdot1.12\simeq0.52\,,\quad r_{ds}^{\prime k^*}[{\rm theo.}]=\frac{\tau_{b_s^*}}{\tau_{b^*}}\cdot1.12\simeq0.52\,.\ ] ] comparing @xmath269 with @xmath3 , one can find that the main difference is induced by the lifetimes of @xmath270 and @xmath271 , _ i.e. _ , @xmath272 but @xmath273 ( see eqs.([eq : gtotbd ] ) and ( [ eq : gtotbs ] ) ) .
all of the findings above for @xmath271 decays are awaiting the future experimental test .
motivated by the rapid development of heavy - flavor experiments , we perform phenomenological studies of nonleptonic @xmath0 and @xmath1 ( @xmath2 ) decays in detail .
the amplitudes are calculated carefully in the framework of qcd factorization , in which , relative to the previous works , the nlo qcd corrections to the transverse amplitudes are evaluated .
our theoretical results for the branching fractions and polarization fractions are summarized in tables [ tab : b ] and [ tab : bstar ] .
after detailed analyses , two possible puzzles relevant to @xmath0 decays are presented , which are : ( i ) the current theoretical results for the ratio @xmath274 , which reflects the effects of flavor - symmetry - breaking and is defined by eq .
( [ eq : rds ] ) , tend to @xmath275 and significantly conflict with the experimental results @xmath276 ; ( ii ) with the form factors extracted precisely from semileptonic @xmath277 decays , the qcdf results for @xmath199 and @xmath200 deviate from the data by about @xmath221 and @xmath222 , respectively .
such tension is also reflected by the form - factor - independent ratio @xmath5 defined by eq .
( [ eq : rvl ] ) , which could be well determined in the sm .
the more experimental and theoretical efforts are required to confirm or refute such two anomalies . for
the @xmath1 decays , they have relatively large branching fractions of the order @xmath6 and are hopeful to be measured by the belle - ii and lhcb experiments .
moreover , they also provide a way for crosschecking of above - mentioned `` @xmath3 and @xmath4 ( or @xmath5 ) puzzles '' through the similar ratios @xmath7 and @xmath8 .
all of the findings in this paper are awaiting the precise test by the refined measurements at lhc and superkekb / belle - ii in the near future .
we thank de - shan yang at ucas , ya - dong yang and xin - qiang li at ccnu for helpful suggestions and discussions . this work is supported by the national natural science foundation of china ( grant nos .
11547014 , 11475055 and 11275057 ) .
q. chang is also supported by the foundation for the author of national excellent doctoral dissertation of p. r. china ( grant no .
201317 ) , the program for science and technology innovation talents in universities of henan province ( grant no .
14hastit036 ) . m. beneke , g. buchalla , m. neubert and c. sachrajda , phys . rev .
* 83 * ( 1999 ) 1914 .
m. beneke , g. buchalla , m. neubert and c. sachrajda , nucl .
b * 591 * ( 2000 ) 313 .
y. y. keum , h. n. li and a. i. sanda , phys .
b * 504 * ( 2001 ) 6 . c. w. bauer and i. w. stewart , phys .
b * 516 * ( 2001 ) 134 . c. w. bauer , d. pirjol and i. w. stewart , phys .
d * 65 * ( 2002 ) 054022 .
m. beneke and m. neubert , nucl .
b * 651 * ( 2003 ) 225 .
m. beneke and m. neubert , nucl .
b * 675 * ( 2003 ) 333 .
m. beneke , j. rohrer and d. s. yang , nucl .
b * 774 * ( 2007 ) 64 .
m. beneke and s. jager , nucl .
b * 751 * ( 2006 ) 160 .
d. s. du , h. j. gong , j. f. sun , d. s. yang and g. h. zhu , phys .
d * 65 * ( 2002 ) 074001 .
j. f. sun , g. h. zhu and d. s. du , phys .
d * 68 * ( 2003 ) 054003 .
h. y. cheng and c. k. chua , phys .
d * 80 * ( 2009 ) 114008 . h. y. cheng and c. k. chua , phys . rev .
d * 80 * ( 2009 ) 114026 . h. y. cheng and k. c. yang , phys .
d * 78 * ( 2008 ) 094001 . | motivated by the rapid development of heavy - flavor experiments , phenomenological studies of nonleptonic @xmath0 and @xmath1 ( @xmath2 ) decays are performed within the framework of qcd factorization .
relative to the previous works , the qcd corrections to the transverse amplitudes are evaluated at next - to - leading order .
the theoretical predictions of the observables are updated . for the measured
@xmath0 decays , the tensions between theoretical results and experimental measurements , _
i.e. _ `` @xmath3 puzzle '' and `` @xmath4 ( or @xmath5 ) puzzle '' , are presented after detailed analyses . for
the @xmath1 decays , they have relatively large branching fractions of the order @xmath6 and are in the scope of belle - ii and lhcb experiments .
moreover , they also provide a way to crosscheck the possible puzzles mentioned above through the similar ratios @xmath7 and @xmath8 .
more refined experimental measurements and theoretical efforts are required to confirm or refute such two anomalies . *
pacs numbers : * 13.25.hw , 12.39.st , 12.38.bx , 14.40.nd |
weakly asymptotically simple and empty ( wase ) space - times provide a setting for the analysis of isolated gravitational objects in classical general relativity .
there is sufficient freedom and generality to allow for the presence of singularities and black holes , but there is also sufficient structure , in the form of a well - behaved asymptotic region , to permit the proof of significant results .
examples of wase space - times include the schwarzschild , kerr and reissner - nordstrm space - times . despite the evident physical importance of wase space - times ,
it is often the case that scant attention is paid to their precise definition . indeed , as will be seen , the definition proposed and used by hawking & ellis @xcite , henceforth h&e , is inadequate for their purposes .
a more restrictive definition was subsequently proposed in @xcite as a foundation for a class of censorship theorems ( see also @xcite ) .
one purpose of the present paper is to provide basic results for this definition further to those given in @xcite .
as has been well - known since @xcite , the concept of a closed trapped surface is central to the understanding of black holes . in the context of wase space - times
it is a standard assertion , stated as proposition 9.2.1 in h&e , that subject to a form of weak cosmic censorship and a suitable energy condition , it is not possible for closed trapped surfaces to be seen from future null infinity . in itself
this is of no direct physical significance because there need be nothing remarkable about the geometry at a space - time point which lies on a closed trapped surface .
the assertion does however imply , again subject to weak cosmic censorship and an energy condition , that the presence of a closed trapped surface in a wase space - time implies the presence of a black hole .
the assertion also leads to an elementary censorship theorem according to which , in a wase satisfying the same energy condition , if every future singularity is sufficiently strong as to be preceded by a closed trapped surface , then weak cosmic censorship must hold .
the second purpose of this paper is to obtain a rigorous proof of a modified statement of h&e proposition 9.2.1 on the basis of the definition of a wase space - time given in @xcite .
the following two basic definitions are taken from @xcite .
[ asymptotedef ] an asymptote of a space - time @xmath1 is a quadruple @xmath2 , where @xmath3 is a space - time - with - boundary , @xmath4 is a @xmath5 real - valued function on @xmath6 , and @xmath7 is a @xmath5 embedding such that a. @xmath8 ; b. @xmath9 ; c. one has @xmath10 and @xmath11 for all @xmath12 .
[ asedef ] an asymptote @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega ' , \psi ' ) $ ] of a space - time @xmath14 is asymptotically simple and empty ( ase ) if a. @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus \supp ( \psi'_*\ricc ( { { \bf g } } ' ) ) \supset \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] ; b. @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] is strongly causal ; c. every inextendible null geodesic @xmath16 of @xmath14 is such that @xmath17 has two endpoints in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] , both of which lie in @xmath18{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
a space - time is asymptotically simple and empty ( ase ) if it admits an asymptotically simple and empty asymptote .
standard arguments give that any ase space - time @xmath1 is globally hyperbolic , and that for any ase asymptote @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega ' , \psi')$ ] of @xmath14 , the boundary @xmath18{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] of @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] is the union of two disjoint connected null hypersurfaces @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+ : = i^+({\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^- : = i^-({\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] . by means of condition ( ii ) of definition [ asedef ]
one can show @xcite that @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] are diffeomorphic to @xmath19 . in the terminology of @xcite , a slice of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ]
is a non - empty locally acausal compact connected topological 2-submanifold of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
theorem 5.1 of @xcite gives that every slice of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] is homeomorphic to @xmath20 .
since , by assumption here , strong causality holds at every point of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] in @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] , proposition 7.1 of @xcite gives that every null geodesic generator of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] cuts every slice of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] , and theorem 7.4 of @xcite gives that @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] is acausal in @xmath3 .
any given slice of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] may be mapped along the generators of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] to yield a foliation of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] by slices of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
similar assertions apply to @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] .
note that @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] need not be causally simple . for example , let @xmath21 be minkowski space .
let @xmath22{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] be a future - directed null geodesic generator of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] and let @xmath23{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] be the antipodal future - directed null geodesic generator of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
let @xmath24 and let @xmath25{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap |{\widetilde{\nu}{\protect \rule{0.0em}{1.0ex}}}|\in { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
one then has @xmath26{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+= { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+\setminus for @xmath27 , for @xmath28 . since @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+\setminus |{\widetilde{\lambda}{\protect \rule{0.0em}{1.6ex}}}_{{\widetilde{q}{\protect \rule{0.0em}{1.0ex}}}}|$ ] is not relatively closed in @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] , the set @xmath26{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] can not be closed in @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] .
condition ( ii ) of definition [ asedef ] may be decomposed into two parts , first that @xmath21 is strongly causal and second that @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] is strongly causal at every point of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] . since the physical interpretation of the latter is unclear one
is led , as in @xcite , to the more general concept of a `` simple '' space - time for which only the chronology condition is imposed on @xmath21 , with no additional causality conditions imposed on @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] or @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] in @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] .
simple space - times are globally hyperbolic with cauchy surfaces which , subject to the truth of the poincar conjecture , are diffeomorphic to @xmath29 .
but the topological and causal structure of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] may exhibit new complications and subtleties . despite the possible interest of this additional generality ,
@xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] will for present purposes be assumed to be strongly causal as expressed in condition ( ii ) if definition [ asedef ] .
in h&e , a weakly asymptotically simple and empty ( wase ) space - time is introduced with a definition which , when re - expressed in terms of asymptotes , assumes the following form .
[ hewase ] a space - time @xmath1 is weakly asymptotically simple and empty ( wase ) if there exists an asymptote @xmath30 of @xmath1 , a space - time @xmath21 , an asymptote @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega ' , \psi ' ) $ ] of @xmath21 and open sets @xmath31 and @xmath32 of @xmath33 and @xmath34 respectively such that a. @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega',\psi ' ) $ ] is ase ; b. @xmath35 is an open neighbourhood of @xmath36 in @xmath6 ; c. @xmath37{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] is an open neighbourhood of @xmath18{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] ; d. @xmath38 and @xmath39 are globally isometric .
this definition of a weakly asymptotically simple and empty space - time is not as restrictive as its authors seem to have intended . in particular
, it allows the future and past null infinities @xmath40 and @xmath41 of @xmath3 to be separated from one another and not join up at a spatial infinity ( see figure 1 ) .
( to add an assumption that @xmath31 is connected would not help . )
the deficiency undermines several of their results , and in particular their proposed proof ( proposition 9.2.1 ) that closed trapped surfaces are necessarily confined to black holes . and
@xmath42 of @xmath1 need not join up at a spatial infinity.,width=384 ] in order to overcome the problems in definition [ hewase ] , one may adopt the following , more restrictive definition of a wase space - time proposed in @xcite .
[ nwase ] an asymptote @xmath30 of a space - time @xmath1 is weakly asymptotically simple and empty ( wase ) if there exists an open set @xmath31 of @xmath33 , an extension @xmath21 of @xmath38 , an asymptote @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega',\psi')$ ] of @xmath21 and a topological embedding @xmath43{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] such that a. @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega',\psi')$ ] is ase ; b. for every @xmath44 , the set @xmath45 is compact ; c. one has @xmath46{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] and @xmath47 and all future - pointing timelike vectors @xmath48 of @xmath1 , the vectors @xmath49 , @xmath50 and @xmath51 are future - pointing in @xmath3 , @xmath14 and @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] respectively .
a space - time is weakly asymptotically simple and empty ( wase ) if it admits a weakly asymptotically simple and empty asymptote . for any cauchy surface @xmath52 of @xmath21 one
has @xmath53 for any @xmath54 and hence that @xmath55 is compact .
the future and past null infinities of a wase asymptote @xmath30 are defined by @xmath40 and @xmath41 respectively .
it is clear that @xmath36 is the disjoint union of @xmath0 and @xmath42 .
the mappings involved in definition [ nwase ] are shown in figure 2 . in figure 3
one can see how the pathologies inherent in definition [ hewase ] are eliminated by condition ( ii ) of definition [ nwase ] .
this condition may be regarded as a way to require that @xmath31 is a neighbourhood of spatial infinity without reference to the geometrical structure of spatial infinity . of a wase space - time .
the left and right squares commute in the category of topological spaces and continuous mappings.,width=288 ] of a wase space - time requires that @xmath31 is such that the darkly shaded region in the right hand figure is compact for all @xmath44 .
this has the desired consequence that @xmath0 and @xmath42 in the left hand figure must meet at a spatial infinity.,width=432 ] the following two lemmas are basic in the analysis of wase space - times .
their proofs are given in @xcite .
[ basiclemma1 ] within the context of definition [ nwase ] one has a. @xmath56{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] is open in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] b. @xmath57 is open in @xmath6 .
[ basiclemma2 ] within the context of definition [ nwase ] one has a. a subset of @xmath31 is open in @xmath33 iff it is open in @xmath34 ; b. a subset of @xmath31 is compact in @xmath33 iff it is compact in @xmath34 ; c. a subset of @xmath31 is closed in @xmath33 if ( but not only if ) it is closed in @xmath34 .
since the mappings @xmath58 and @xmath59{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] in definition [ asymptotedef ] are both diffeomorphisms onto their images , it is clear from figure 2 that the mapping @xmath60{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] in definition [ nwase ] is such that @xmath61 is a diffeomorphism onto @xmath62 .
moreover @xmath63 is a conformal isometry onto @xmath62 in the sense of @xmath64 .
note however that @xmath65 need not be differentiable at points of @xmath36 . [ generators ] one has a. @xmath66{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] and @xmath67{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] ; b. @xmath65 maps the null geodesic generators of @xmath0 and @xmath42 onto null geodesic generators of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] respectively .
let @xmath68 .
lemma [ basiclemma1 ] gives that @xmath69 is an open neighbourhood of @xmath70 in @xmath6 .
hence , by the definition of @xmath0 , there exists a smooth timelike curve @xmath71 of @xmath3 having a future endpoint at @xmath70 in @xmath3 .
one has @xmath72{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] by condition ( iii ) of definition [ nwase ] . since @xmath73 is a smooth timelike curve of @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } }
{ \protect \rule{0.0em}{1.0ex}}})$ ] with a future endpoint at @xmath74 one therefore has @xmath75{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] by the definition of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
there follows @xmath76{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
a similar argument gives @xmath77{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+ ) \subset { \mathscr{i}}^+$ ] which implies @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+ \subset \xi ( { \mathscr{i}}^+)$ ]
. hence one has @xmath66{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] and similarly @xmath67{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] .
this establishes ( i ) .
let @xmath78 be a future - directed null geodesic generator of @xmath0 and let @xmath79 .
let @xmath80\subset i$ ] for @xmath81 and let @xmath82 .
one has @xmath83 . by lemma [ basiclemma1 ] and the definition of @xmath0
there exists a smooth future - directed timelike curve @xmath84 of @xmath3 with a future endpoint at @xmath85 .
indeed there exists a sequence of smooth future - directed timelike curves @xmath86 of @xmath3 converging pointwise to @xmath87 in @xmath6 , each with a future endpoint at @xmath88 .
since the smooth timelike curves @xmath89 of @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] converge pointwise to @xmath90 in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] one has that @xmath91 is a causal curve of @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] with a future endpoint at @xmath92 .
since part ( i ) gives @xmath93{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] and since @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] is a null hypersurface of @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] it follows that @xmath94 is a null geodesic generating segment of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
since @xmath95 and @xmath96 were arbitrary in @xmath82 it follows that @xmath97 is a null geodesic generating segment of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
clearly @xmath98 can not have a future endpoint @xmath99{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] otherwise @xmath100 would have a future endpoint at @xmath101 and so would not be a generator of @xmath0 , contrary to hypothesis . similarly @xmath98 can not have a past endpoint in @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
hence @xmath98 is a null geodesic generator of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
the corresponding result for @xmath42 is similar .
this establishes ( ii ) .
proposition [ generators ] shows that the structure of @xmath0 and @xmath42 for a wase asymptote @xmath102 is directly analogous to the structure of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] for an ase asymptote @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega',\psi)$ ] . in particular one may , following @xcite , define a slice of @xmath0 ( respectively a slice of @xmath42 ) as a non - empty locally acausal compact connected topological 2-submanifold of @xmath0 ( respectively @xmath42 )
. then @xmath0 and @xmath42 are acausal in @xmath3 , every null geodesic generator of @xmath0 cuts every slice of @xmath0 and every null geodesic generator of @xmath42 cuts every slice of @xmath42 .
slices of @xmath0 and @xmath42 are mapped by @xmath65 to slices of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] respectively .
slices of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] are mapped by @xmath103 to slices of @xmath0 and @xmath42 respectively .
the following is a useful restriction on the causal structure of a wase space - time .
it is equivalent to a definition of asymptotic simplicity in @xcite but is re - expressed here in a form more convenient for present purposes .
the change of terminology seems appropriate because the term `` simple '' has become overworked .
[ acc ] a wase asymptote @xmath30 is asymptotically chronologically consistent if @xmath31 , @xmath21 and @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega',\psi')$ ] in definition [ nwase ] may be chosen such that for any achronal set @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{a}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$ ] of @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] such that @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{a}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}\subset \psi'({\mathscr{u}})\cup \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ]
one has that @xmath77{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{a}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}})\subset \psi ( { \mathscr{u}})\cup \partial { \bar{m}{\protect \rule{0.0em}{1.6ex}}}$ ] is achronal in @xmath3 .
a form of weak cosmic censorship hypothesis will be required .
the h&e concept of future asymptotic predictability is suitable for this purpose .
the following definition formulates future asymptotic predictability in terms of asymptotes and provides a weaker concept of partial future asymptotic predictability that is also well - established in the literature .
let @xmath104 be a closed achronal set without edge in a wase space - time @xmath1 and let @xmath102 be a wase asymptote of @xmath1 . then @xmath102 is future asymptotically predictable from @xmath105 if one has @xmath106 .
one says that @xmath102 is partially future asymptotically predictable from @xmath105 if there exists a slice @xmath107 of @xmath0 such that @xmath108 .
the main result is the following : [ maintheorem ] let @xmath3 be a wase space - time and let @xmath102 be a wase asymptote of @xmath1 .
suppose 1 .
there exists a closed , edgeless achronal set @xmath104 in @xmath1 such that @xmath102 is future asymptotically predictable from @xmath105 ; 2 .
@xmath102 is asymptotically chronologically consistent ; 3 .
one has @xmath109 for all null vectors @xmath110
, then for any closed trapped surface @xmath111 of @xmath1 in @xmath112 one has @xmath113 . conditions ( 1 ) and ( 3 ) of theorem [ maintheorem ] coincide with conditions in the statement of h&e proposition 9.2.1 . however condition ( 2 ) , which would seem to be necessary , makes no appearance in h&e .
note also that theorem [ maintheorem ] requires only @xmath114 whereas h&e impose the stronger condition @xmath115 .
the basic idea of the h&e argument in support of the statement of their proposition 9.2.1 is to show that if @xmath111 is visible from @xmath0 then there must be a null geodesic generator of @xmath116 which reaches from @xmath111 to @xmath0 and which is therefore of infinite affine length .
a contradiction then follows by means of the raychaudhuri equation and the null convergence condition .
the argument fails though because the h&e definition of a wase space - time is not sufficiently strong .
specifically , things begin to go wrong when they claim that , in the associated ase space - time @xmath21 , for a cauchy surface @xmath52 chosen such that @xmath117 it is necessarily the case that @xmath118 is compact .
( at this point h&e are tacitly identifying @xmath31 and @xmath32 , as is explicitly done in the present formalism . ) in the first place it is unclear that there need be any cauchy surface @xmath52 of @xmath21 such that @xmath119 .
for example @xmath120 might not be achronal in @xmath21 . and
second , since @xmath104 need not even intersect @xmath31 and every cauchy surface of @xmath21 is non - compact , the set @xmath121 could be non - compact .
definition [ nwase ] directly overcomes the second difficulty , as was indicated in the remark that followed definition [ nwase ] . in order to overcome the first it will be necessary to employ different techniques . .
the closed trapped surface @xmath111 is visible from @xmath0 but no null geodesic generator of the boundary of the causal future of @xmath111 meets @xmath0.,width=336 ] at a pictorial level one might seek a counterexample to h&e proposition 9.2.1 by arranging to have @xmath0 both separated from @xmath42 and contained in the chronological future of a closed trapped surface @xmath111 ( see figure 4 ) . since none of the null geodesic generators of @xmath122 then meet @xmath0 , the central contradiction in the proof of h&e lemma 9.2.1 is avoided .
however to obtain a full counterexample to h&e proposition 9.2.1 one also needs to arrange for the null convergence condition to be satisfied .
even though it is not clear how this might be done , it seems unlikely that h&e proposition 9.2.1 is correct .
the following three lemmas and a proposition are the key to the proof of the revised trapped surfaces theorem .
[ mainlemma ] let @xmath1 be a wase space - time with a asymptotically chronologically consistent wase asymptote @xmath102 and let @xmath123 be a slice of @xmath42 .
then there exists a slice @xmath107 of @xmath0 such that @xmath124 is a closed set of @xmath6 contained in @xmath69 and @xmath125{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is a closed set of @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] contained in @xmath126{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] for @xmath127 and @xmath128 .
the time reverse of lemma 3.6 of @xcite gives that @xmath129 can not be contained entirely in @xmath130{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] .
let @xmath131{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus ( i^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex } } } ) \cup \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] and let @xmath132{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\setminus \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
one then has @xmath133{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus ( j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex } } } ) \cup \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] .
let @xmath134 .
the set @xmath135 is compact in @xmath34 by definition [ nwase ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{q}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}:=\psi'({\mathscr{q } } ' ) $ ] is compact in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
so by lemma 4.5 ( iii ) of @xcite the set @xmath136{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{q}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] is a slice of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
the set @xmath137{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\setminus ( \psi'({\mathscr{u}})\cup \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is contained in @xmath138{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] which does not intersect @xmath139{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] . hence @xmath137{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is contained in @xmath126{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
the set @xmath140 is a slice of @xmath0 .
suppose there exists a point @xmath141 .
then there exists a causal curve @xmath142 of @xmath3 from @xmath123 to @xmath143 , and a timelike curve @xmath144\rightarrow { \bar{m}{\protect \rule{0.0em}{1.6ex}}}$ ] of @xmath3 from @xmath143 to @xmath145 .
let @xmath146 : \beta ( t)\notin \psi ( { \mathscr{u}})\cup \partial { \bar{m}{\protect \rule{0.0em}{1.6ex}}}\}$ ] .
then @xmath147 lies in the topological boundary of @xmath57 in @xmath6 and one has @xmath148 for all @xmath149 $ ] . for each @xmath150 $ ] the set
@xmath151 is non - achronal in @xmath3 and contained in @xmath69 .
so , by the asymptotic chronological consistency of @xmath102 , one has that @xmath152 is non - achronal in @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] for each @xmath149 $ ] . since @xmath153\rightarrow { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] is a timelike curve of @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] from @xmath154 to @xmath155 one thus has @xmath156{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_0,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] for all @xmath157 $ ] .
lemma 4.12 ( i ) of @xcite gives that @xmath139{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\supset i^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is closed in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
hence one has @xmath158{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^-(\sigma^+_0,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] for all @xmath159 $ ] .
but since @xmath147 lies in the topological boundary of @xmath160 in @xmath6 the point @xmath161 must lie in the topological boundary of @xmath126{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
this is impossible because @xmath137{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is contained in @xmath126{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] which is open in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
one thus has @xmath162 .
now let @xmath163 be a slice of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] lying strictly to the past of @xmath155 along the generators of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
one has @xmath125{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect
\rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset j^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_0,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset \psi'({\mathscr{u}})\cup \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
hence a point of @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] lies in @xmath125{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] iff it lies on a causal curve of @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] in @xmath164{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] from @xmath165 to @xmath163 .
moreover , since lemma 4.12 ( i ) of @xcite gives that @xmath137{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] and @xmath139{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] are both closed in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] , the set @xmath125{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is closed in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
let @xmath166 .
one has @xmath167 .
hence a point of @xmath6 lies in @xmath124 iff it lies on a causal curve of @xmath3 in @xmath69 from @xmath123 to @xmath107 .
one thus has @xmath168{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+(\sigma^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}))$ ] .
since @xmath125{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] has been shown to be closed in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] it follows that @xmath169 is closed in @xmath6 .
let @xmath170 be a compact set of @xmath33 .
then there exists a slice @xmath171 of @xmath0 such that @xmath172 does not intersect @xmath173 .
one may assume @xmath174 otherwise one may redefine @xmath31 as @xmath175 .
the lemma then gives that there exists a slice @xmath171 of @xmath0 such that @xmath176 .
[ secondlemma ] let @xmath123 be a slice of @xmath42 and let @xmath177 be an open neighbourhood of @xmath123 in @xmath6 .
then there exists a slice @xmath178 of @xmath0 such that , for every @xmath179 , the set @xmath177 is cut by every past endless timelike curve of @xmath3 to @xmath180 .
it suffices to assume @xmath181 since one may otherwise redefine @xmath177 as @xmath182 .
the set @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{n}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}:=\xi({\mathscr{n}})$ ] is open in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] and @xmath183 is a slice of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] . by the time reverse of lemma 4.12 of @xcite one
has that @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is compact in @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] and such that @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-={\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-$ ] .
the time reverse of proposition 7.2 of @xcite gives @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+\subset i^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] whereby one has @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+=\emptyset$ ] .
hence one has @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect
\rule{0.0em}{1.6ex}}})\cap \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}={\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-$ ] .
let @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{x}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$ ] be the compact set @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\setminus { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{n}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}\subset { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] or , if this is empty , let @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{x}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$ ] be any non - empty compact set of @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] . the set @xmath185{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{x}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$}}:= { \dot{j}{\protect \rule{0.0em}{1.6ex}}}^+({\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{x}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] is a slice of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] . since @xmath186{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{x}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] does not intersect @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\setminus { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{n}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$ ] one has @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect
\rule{0.0em}{1.6ex}}})\cap i^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_{\mbox{\scriptsize${\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{x}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{n}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$ ] .
let @xmath107 be as in the statement of lemma [ mainlemma ] and let @xmath128 .
let @xmath187 be a slice of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] lying strictly to the past of both @xmath185{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{x}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$}}$ ] and @xmath163 along the generators of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+$ ] .
one then has @xmath188{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset \psi'({\mathscr{u}})\cup \partial { \bar{m}{\protect \rule{0.0em}{1.6ex}}}$ ] and @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_2,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset { \dot{j}{\protect \rule{0.0em}{1.6ex}}}^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap ( i^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_{\mbox{\scriptsize${\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{x}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cup { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+ ) \subset { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{n}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$ ] . the set @xmath189 is a slice of @xmath0 .
let @xmath179 and let @xmath190\rightarrow { \bar{m}{\protect \rule{0.0em}{1.6ex}}}\setminus \partial { \bar{m}{\protect \rule{0.0em}{1.6ex}}}$ ] be a future - directed , past endless timelike curve of @xmath3 having a future endpoint at @xmath191 .
let @xmath192\rightarrow { \bar{m}{\protect \rule{0.0em}{1.6ex}}}$ ] be the maximal segment of @xmath193 to @xmath194 in @xmath69 .
then @xmath195 is a timelike curve of @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] to @xmath196 .
one clearly has @xmath197{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] . in order to show that @xmath198 cuts @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{n}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}$ ]
it therefore suffices to show that @xmath198 cuts @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] .
suppose first that @xmath198 is past endless in @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] .
the time reverse of lemma 3.6 of @xcite gives that @xmath130{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] can not contain all of @xmath199{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] , whilst the time reverse of lemma 4.2 of @xcite gives @xmath200{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] .
hence @xmath198 cuts @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus i^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] .
since @xmath198 is past endless and timelike in @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] it follows that @xmath198 cuts @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] . because @xmath198 has a future endpoint at @xmath201{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+\subset i^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ]
one thus has that @xmath198 cuts both @xmath130{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex } } } ) \subset j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] and so cuts @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] .
now suppose that @xmath198 has a past endpoint @xmath202 in @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] .
the point @xmath202 must lie in the topological boundary of the open set @xmath126{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] otherwise it would lie in @xmath126{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] , in which case @xmath203 would be a past endpoint to @xmath204 in @xmath1 and @xmath204 would be past extendible in @xmath69 . because @xmath126{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] is open in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] the set @xmath125{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset \psi'({\mathscr{u}})\cup \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] can not contain @xmath202 . since @xmath125{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is closed in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] the set @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus ( j^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}))$ ] is an open neighbourhood of @xmath202 in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] and so is cut by @xmath198 . in view of @xmath197{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset j^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] one thus has that @xmath198 cuts @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] . since @xmath198 has a future endpoint at @xmath205{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^+\subset i^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] it follows that @xmath198 cuts both @xmath130{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] and @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}\setminus j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] and so cuts @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] . since @xmath195 cuts @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] and therefore cuts @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{n}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}:=\xi({\mathscr{n}})$ ] one has that @xmath204 cuts @xmath177 .
hence @xmath193 cuts @xmath177 .
the final lemma will require the use of the following result .
[ closedprop ] let @xmath102 be a wase asymptote of a wase space - time @xmath1 and let @xmath104 be a closed edgeless achronal set of @xmath1
. if @xmath102 is partially future asymptotically predictable from @xmath105 then @xmath105 is closed edgeless and achronal in @xmath3 .
let @xmath21 be the associated ase space - time and @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega',\psi')$ ] an ase asymptote of @xmath21 . if @xmath206 was a timelike curve of @xmath3 from @xmath207 to @xmath207 then , because @xmath36 is a null hypersurface of @xmath3 , one would have @xmath208 so there would exist a timelike curve @xmath209 of @xmath1 such that @xmath210 .
but then @xmath209 would be a timelike curve of @xmath1 from @xmath104 to @xmath104 .
this would contradict the achronality of @xmath104 in @xmath1 . hence @xmath105 is achronal in @xmath3 .
suppose there exists @xmath211 .
since @xmath102 is partially future asymptotically predictable from @xmath105 there exists a slice @xmath107 of @xmath0 such that @xmath212 . in the case
@xmath213 , every past endless timelike curve of @xmath3 to @xmath214 would cut @xmath105 , and so @xmath215 would be an open neighbourhood of @xmath216 and so would intersect @xmath105 .
this is impossible since @xmath105 is achronal in @xmath3 .
so suppose @xmath217 .
there exists @xmath218 lying strictly to the past of @xmath214 on the null geodesic generator of @xmath0 through @xmath214 . in view of @xmath219 , every past endless timelike curve of @xmath3 to @xmath220 must cut @xmath105 .
but then @xmath215 is an open neighbourhood in @xmath6 of @xmath221 and therefore of @xmath222 and so must intersect @xmath105 .
so again one has a contradiction to the achronality of @xmath105 .
one thus has @xmath223 .
suppose there exists @xmath224 .
let @xmath123 be a slice of @xmath42 such that @xmath225 .
by lemma [ mainlemma ] there exists a slice @xmath107 of @xmath0 such that @xmath125{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset \psi'({\mathscr{u}})\cup \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] for @xmath127 and @xmath128 .
since @xmath102 is partially future asymptotically predictable from @xmath105 one may assume that @xmath107 is taken sufficiently far to the past in @xmath0 to give @xmath226 .
let @xmath227 . by lemma 7.2 of @xcite one
\rule{0.0em}{1.6ex}}}^-\subset i^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] .
hence there is a timelike curve @xmath228 of @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}})$ ] from @xmath229 to some point @xmath230 . in view of @xmath231{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\subset \psi'({\mathscr{u}})\cup \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] one has that @xmath232 is a timelike curve of @xmath3 from @xmath224 to @xmath233 .
the set @xmath234 can not intersect @xmath105 otherwise @xmath235 would be an open neighbourhood of @xmath236 and so @xmath105 would not be achronal in @xmath3 . since the past endless null geodesic generating segment of @xmath42 to @xmath70 clearly does not cut @xmath105
it follows that @xmath234 does not intersect @xmath237 .
but @xmath234 is a neighbourhood of @xmath238 in @xmath6 so one has a contradiction .
one now has @xmath239 . since @xmath58 is a diffeomorphism onto its image it follows that @xmath105 is relatively closed in @xmath240 .
hence @xmath105 is closed in @xmath6 .
one has @xmath241 because @xmath105 is closed in @xmath6 and does not intersect @xmath36 . and , because @xmath242 is a conformal isometry onto its image , one has @xmath243 .
hence @xmath244 is empty .
[ lemma3 ] let @xmath3 be a wase space - time and let @xmath102 be a wase asymptote of @xmath1 .
suppose 1 .
there exists a closed edgeless achronal set @xmath104 in @xmath1 such that @xmath102 is partially future asymptotically predictable from @xmath105 ; 2 .
@xmath102 is asymptotically chronologically consistent .
then for any compact set @xmath245 of @xmath33 there exists a slice @xmath246 of @xmath0 such that @xmath247 .
it suffices to assume @xmath248 since one may otherwise redefine @xmath31 as @xmath175 .
let @xmath249 be a slice of @xmath42 and let @xmath123 be a slice of @xmath42 lying strictly to the past of @xmath249 along the null geodesic generators of @xmath42 .
let @xmath250 and @xmath127 .
by lemma [ mainlemma ] there exists a slice @xmath145 of @xmath0 such that @xmath251 and @xmath139{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_0,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\bar{m}{\protect \rule{0.0em}{1.6ex}}})\subset \psi'({\mathscr{u}})\cup \partial { \makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] for @xmath252 .
lemma 4.12 of @xcite gives @xmath184{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap { \mathscr{i}}^-={\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-$ ] which , since @xmath253 lies strictly to the future of @xmath165 along the null geodesic generators of @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-$ ] , implies that @xmath139{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is a neighbourhood of @xmath253 in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
proposition 7.2 of @xcite gives @xmath254{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{i}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}^-\subset i^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_0,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] whereby one has that @xmath137{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\supset i^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_0,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] is a neighbourhood of @xmath253 in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
thus there exists an open neighbourhood @xmath15{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{n}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}}\subset j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_0,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})$ ] of @xmath253 in @xmath15{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}$ ] .
the set @xmath255{\hspace{1pt}$\widetilde{\hspace{-3.0pt}\makebox[0.9em][r]{$\mathscr{n}$}}$\hspace{-2pt}}{\protect \rule{0.0em}{1.6ex}}})\subset \xi^{-1}(j^+({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^-,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}})\cap j^-({\widetilde{\sigma}{\protect \rule{0.0em}{1.6ex}}}^+_0,{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}};{\makebox[1em]{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}}))= j^+(\sigma^-,{\bar{{{\bf g}}}{\protect \rule{0.0em}{1.0ex}}};{\bar{m}{\protect \rule{0.0em}{1.6ex}}})\cap j^-(\sigma^+_0,{\bar{{{\bf g}}}{\protect \rule{0.0em}{1.0ex}}};{\bar{m}{\protect \rule{0.0em}{1.6ex}}})$ ] is an open neighbourhood of @xmath249 in @xmath6 . in view of proposition [ closedprop ]
one may , by passing to a subset of @xmath177 if necessary , assume @xmath256 . by passing to a further subset of @xmath177 if necessary
, one may arrange that each point of @xmath257 is a future endpoint of a timelike curve of @xmath3 in @xmath177 from @xmath258 .
lemma [ secondlemma ] gives that there exists a slice @xmath178 of @xmath0 such that every past endless timelike curve of @xmath259 to @xmath178 cuts @xmath177 .
one may assume that @xmath178 lies strictly to the past of @xmath145 along the null geodesic generators of @xmath0 .
since @xmath102 is partially future asymptotically predictable from @xmath105 there exists a slice @xmath246 of @xmath0 lying strictly to the past of @xmath178 along the generators of @xmath0 such that @xmath260 .
suppose there exists a timelike curve @xmath261\rightarrow { \bar{m}{\protect \rule{0.0em}{1.6ex}}}$ ] of @xmath3 from @xmath262 to @xmath246 .
in view of @xmath263 the set @xmath264 can not intersect @xmath265 .
so , because every past endless timelike curve of @xmath3 to @xmath246 cuts @xmath266 , there exists @xmath267 such that @xmath268 . by the construction of @xmath177 there exists @xmath269 .
if @xmath270 $ ] did not cut @xmath105 one could concatenate the past endless null geodesic generating segment of @xmath271 to @xmath272 , a timelike curve in @xmath273 from @xmath272 to @xmath274 , and the segment @xmath275 $ ] of @xmath142 from @xmath276 to @xmath277 to obtain a past endless causal curve @xmath278 of @xmath3 to @xmath279 which did not cut @xmath105 . for
an open neighbourhood @xmath280 of @xmath281 in @xmath6 there would exist @xmath282 such that @xmath283 .
but then @xmath284 would be an open neighbourhood of @xmath285 in @xmath6 not intersecting @xmath237 , which gives a contradiction .
thus @xmath270 $ ] must cut @xmath105 and indeed there must exist @xmath286 such that @xmath287 .
one now has that @xmath288 $ ] is a timelike curve of @xmath3 from @xmath289 to @xmath287 .
hence @xmath235 intersects @xmath290 and so intersects @xmath215 .
this contradicts the achronality of @xmath105 in @xmath3 .
hence there can be no timelike curve of @xmath3 from @xmath173 to @xmath246 .
one thus has @xmath291 .
suppose there exists @xmath292 .
since @xmath293 is relatively open in @xmath0 one can construct an open neighbourhood @xmath294 of @xmath295 in @xmath6 such that every point of @xmath296 is a past endpoint of a timelike curve of @xmath3 in @xmath294 to @xmath297 .
then @xmath298 intersects @xmath299 and so @xmath300 intersects @xmath173 , which is impossible .
hence @xmath301 does not intersect @xmath293 .
there follows @xmath302 .
it is now possible to give the proof of theorem [ maintheorem ] .
one may , by passing to a subset of @xmath31 if necessary , assume @xmath303 .
suppose , for the purpose of obtaining a contradiction , that @xmath304 is non - empty .
then @xmath305 is non - empty and so is @xmath306 . by lemma [ lemma3 ]
there exists a slice @xmath107 of @xmath0 such that @xmath307 .
since @xmath308 is a non - empty proper subset of @xmath0 it follows that @xmath309 is a non - empty proper subset of @xmath0 .
hence there exists @xmath310 .
there exists a null geodesic generator @xmath311 of @xmath312 to @xmath313 having either a past endpoint in @xmath314 or no past endpoint in @xmath6 . in the former case
@xmath311 could not be a null geodesic generating segment of @xmath0 because it would have a past endpoint in @xmath315 . in the latter case
@xmath311 could not be a null geodesic generating segment of @xmath0 because it would then cut @xmath316 .
hence one has @xmath317 .
suppose @xmath311 were past endless in @xmath3
. then @xmath311 would be a past endless causal curve of @xmath3 in @xmath312 .
let @xmath318 lie strictly to the future of @xmath313 along the null geodesic generator of @xmath0 though @xmath313
. then one could deform @xmath311 to the future in @xmath3 so as to give a past endless timelike curve @xmath319 of @xmath3 to @xmath320 in @xmath321 . clearly @xmath319 could not intersect @xmath105 because @xmath105 is achronal in @xmath3 .
since @xmath319 is timelike curve of @xmath3 to @xmath322 in @xmath3 it follows that there would exist a neighbourhood of @xmath320 in @xmath6 that did not intersect @xmath237 .
this would be contrary to the future asymptotic predictability of @xmath102 from @xmath105 .
thus @xmath311 must have a past endpoint in @xmath3 at @xmath314 .
consequently there exists a null geodesic @xmath100 of @xmath1 such that @xmath323 is the unique maximal segment of @xmath311 in @xmath129 . for each @xmath324 , every open neighbourhood of @xmath325 in @xmath6 intersects both @xmath326 and @xmath327 .
one thus has @xmath328 and hence that @xmath100 is a null geodesic generator of @xmath116 .
since @xmath311 has a past endpoint at @xmath314 in @xmath3 it follows that @xmath100 has a past endpoint at @xmath111 in @xmath1 .
since @xmath311 has a future endpoint at @xmath329 in @xmath3 it follows that @xmath100 is future endless and future complete in @xmath1 .
one may assume that @xmath100 is an affine future - directed null geodesic of @xmath1 of the form @xmath330 .
let @xmath331 and @xmath332 be null normal fields to @xmath111 along a relative open neighbourhood @xmath333 of @xmath334 in @xmath111 , normalised such that @xmath335 , with @xmath336 .
the induced metric on @xmath111 is given by @xmath337 , whilst @xmath338 and @xmath339 are null second fundamental forms of @xmath111 along @xmath340 . by the definition of a closed trapped surface one has @xmath341 and @xmath342 along @xmath333 .
the vector field @xmath331 along @xmath340 defines a congruence of future endless affine null geodesics of @xmath1 from @xmath111 with tangents that coincide with @xmath331 along @xmath340 .
let @xmath331 also denote the tangents to these null geodesics .
each tangent vector to @xmath111 at @xmath334 may be lie propagated along @xmath100 with respect to @xmath331 to yield a vector field @xmath343 along @xmath100 . from vanishing torsion
one has @xmath344 and hence that @xmath343 satisfies the defining equation @xmath345 for a jacobi field along @xmath100 .
note that @xmath343 is orthogonal to @xmath331 at @xmath346 and satisfies @xmath347 along @xmath100 , and so is orthogonal to @xmath331 along @xmath100 .
one may parallelly propagate the vector @xmath332 along the integral curves of @xmath331 and so define @xmath337 along these curves .
then @xmath348 is a projection operator such that @xmath349 .
one thus has that @xmath350 satisfies @xmath351 for @xmath352 now defined all along @xmath100 .
one also has @xmath353 .
the expansion and shear tensors of the vector fields @xmath354 along @xmath100 may be expressed as @xmath355 and @xmath356 respectively , where @xmath357 is the scalar expansion .
then , defining @xmath358 , one has that @xmath359 satisfies the raychaudhuri equation @xmath360 by means of condition ( 3 ) of theorem [ maintheorem ] one thus has @xmath361 for all @xmath362 such that @xmath363 .
hence there exists @xmath364 $ ] such that @xmath365 .
thus @xmath366 is conjugate to @xmath111 at @xmath367 and so there exists a jacobi field @xmath343 along @xmath100 which is non - zero and tangent to @xmath111 at @xmath368 and such that @xmath369 .
one may , by an adaptation of the technique of h&e proposition 4.5.12 , use the vector field @xmath354 to construct a timelike curve of @xmath1 from @xmath111 to @xmath370 for any @xmath371 .
but this is impossible because @xmath100 is a generator of @xmath116 .
this establishes the required contradiction .
this then is the revised proof of the familiar assertion that , subject to the null convergence condition and weak cosmic censorship , closed trapped surfaces are not visible from @xmath0 .
the key idea of the h&e argument in support of this assertion was evidently sound , but additional constraints and analysis have been seen to be necessary to make the detailed theory of wase space - times match intuitive expectations .
one could , as discussed in [ prelimsec ] , consider weakening the assumed strong causality of the reference ase asymptote @xmath13{\hspace{1pt}$\widetilde{\hspace{0.5pt}\makebox[0.9em][r]{$m$}}$}{\protect \rule{0.0em}{1.6ex}}},{\widetilde{{{\bf g } } } { \protect \rule{0.0em}{1.0ex}}},\omega',\psi')$ ] to a the chronology condition on the underlying ase space - time @xmath21 .
this would lead to a more general definition of a wase space - time .
it would be of some interest to find whether or not theorem [ maintheorem ] remains true in this more general setting .
the definition of a wase space - time , proposed in @xcite as the foundation for certain types of cosmic censorship theorems , has been seen also to provide the basis for a rigorous proof of standard relativity folklore concerning the invisibility of closed trapped surfaces from @xmath0 .
other approaches to the definition of wase space - time have been attempted @xcite , but a comparison will not be attempted here . whilst this paper was in preparation ,
chruciel et al.@xcite have been reconsidering another piece of relativity folklore , namely the area theorem for black holes .
this states that , subject to weak cosmic censorship and an energy condition , the area of a black hole can not decrease .
a formalised statement to this effect appears as proposition 9.2.7 in h&e , but the proof there is flawed because it is based on unsubstantiated assumptions concerning the smoothness of the event horizon . a proof is provided in @xcite of a weaker area theorem , but with full attention to matters of differentiability .
a central hypothesis is one of @xmath372-regularity which , in present terminology , requires that there exists a neighbourhood @xmath373 of the event horizon @xmath374 in @xmath6 such that , for any compact set @xmath375 which intersects @xmath376 , there is a null geodesic generator of @xmath0 which cuts both @xmath377 and @xmath378 .
the authors consider various ways to derive @xmath372-regularity from seemingly more natural hypotheses .
however , lemma [ lemma3 ] of the present paper shows that for space - times which are wase in the sense of @xcite and asymptotically chronologically consistent in the sense of definition [ acc ] , @xmath372-regularity is in fact a consequence of partial future asymptotic predictability .
i am grateful to piotr chruciel for having prompted me to write this article , and for his subsequent comments .
financial support was provided by the nfr ( natural sciences research council ) of sweden .
censorship , persistent curvature and asymptotic causal pathology . in j.n .
islam w.b .
bonnor and m.a.h .
maccallum , editors , _ classical general relativity _ , pages 193207 .
cambridge , 1984 . | it is standard assertion in relativity that , subject to an energy condition and the cosmic censorship hypothesis , closed trapped surfaces are not visible from @xmath0 . a proof given by hawking &
ellis in _ the large scale structure of space - time _ is flawed since it is formulated in terms of an inadequate definition of a weakly asymptotically simple and empty space - time . a new proof is given based on a more restrictive definition of a weakly asymptotically simple and empty space - time . |
exploring the use of individual molecules as active components in electronic devices @xcite has been at the forefront of nanoelectronics research in recent years due to the potential advantage of ultrahigh density / speed and low - cost device fabrication through self - assembly / self - organization processes that such molecular - scale devices may bring @xcite .
numerous useful device characteristics including molecular rectifying diodes , negative differential resistance and field - effect transistors have been demonstrated using molecular - scale structures including small conjugated molecules @xcite , single- and multi - wall carbon nanotubes @xcite and macromolecules like dna @xcite . a quantitative understanding of the physical mechanisms underlying the operation of such diverse molecular - scale devices has been and remains a major challenge in nanoelectronics research .
the electrical characteristics of molecular devices are usually measured by sandwiching the molecules between two metallic electrodes @xcite .
the measured transport characteristics reflects both the nature of the metal - molecule interface and the properties of the molecular layer @xcite .
there are basically three electronic processes of interest in such metal - molecule - metal junctions : charge transfer between the metals and the molecules , the change of the electrostatic potential and the modification of the molecular geometry and electronic states .
the nature of these processes under both zero and nonzero bias determines the electrical characteristics of the molecular junction .
here it is natural to separate the problem into device at equilibrium and device out of equilibrium since : ( 1 ) the linear dc - transport property of the molecular device probes the equilibrium charge distribution of the molecular junction established by adsorption onto the electrodes ; ( 2 ) the nonlinear transport probes the charge response of the molecular junction to the applied electrical field established through a finite bias voltage .
a microscopic theory of molecular electronics will therefore need to : ( 1 ) determine the appropriate geometry of the molecular junction ; ( 2 ) determine the self - consistent charge transfer and the resulting lineup of the molecular levels relative to the metal fermi - level and the modification of the molecular states upon formation of the metal - molecule - metal junction ; ( 3 ) determine the molecular screening of the applied electric field , the field - induced modification of molecular states ( the static stark effect ) and the non - equilibrium electron distribution when current is flowing ; ( 4 ) determine the current / conductance - voltage characteristics and their correlation with the molecular and device structures .
none of these issues has been satisfactorily solved such that theory can be used in conjunction with experiment for an unambiguous identification of the device operation mechanism .
the purpose of our work is to elucidate the above device electronic processes and their dependence on the given molecular and device structures under both equilibrium and nonequilibrium conditions within a well - defined theoretical model .
specifically , we will investigate the different aspects of molecular transport due to the interplay between the electronic processes at the metal - molecule interface , the electronic processses in the molecule core and the molecular response to the applied electrical field when current is flowing .
correspondingly , we focus our attention on the electron dynamics and calculate the device characteristics within the coherent transport regime .
our emphasis is on the conceptual understanding and the chemical trends obtained from detailed microscopic calculation of simple but representaive molecular and device structures .
our work can therefore be considered as a `` minimal '' quantitative microscopic model of single - molecule electronics .
the ionic dynamics of the metal - molecule - metal junction can also be affected by the above electronic processes both at equilibrium and out of equilibrium , i.e. , the adsorption and bias / current - induced conformation change .
neither aspect will be treated here . by confining ourselves to the coherent transport regime
, we also neglect the effect of electron - vibrational coupling on transport @xcite .
solving the adsorption - induced conformation change requires an accurate knowledge of the surface lattice structure at the atomic - scale for devices under applied voltage .
this is almost never known in typical molecular transport measurement .
since one goal of such calculation is to provide the input nuclear configuration for transport calculation , our purpose can be served equally well by examining the effect that different adsorption and molecular geometries may have on the device characteristics , which will be treated in subsequent papers of the series .
a rigorous solution of the bias / current - induced conformational change and the general solution of inelastic contribution to non - linear current due to electron - vibrational coupling is a challenging topics which requires solving the nonequilibrium dynamics of the coupled electron - nuclei systems @xcite since the molecular potential energy surface is affected by the nonequilibrium electron dynamics and electron energy is dissipated in moving the atoms . by focusing on the electron dynamics in the molecular junction with fixed geometry
, the present work will provide a reference for evaluating the importance of such additional complications and provide the necessary input for further investigation in situations where they are indeed critical . in typical molecular junction measurement ,
conjugated molecules are attached to the metallic electrodes through appropriate end groups . in this paper
we will consider current transport through two benene - based molecular radicals
phenylene dithiol ( pdt ) and biphenylene dithiol ( bpd)adsorbed symmetrically onto two semi - inifinite gold @xmath1111@xmath2 electrodes through the end sulfur atoms .
these structures chosen are among the simplest possible but are still representative of current experimental work . for the device at equilibrium , we obtain the self - consistent charge transfer , the adsorption - induced change in the electrostatic potential , the lineup of the molecular level relative to the metal fermi - level and the modification of the individual molecular states due to the metal - molecule coupling . for the device out of equilibrium ,
we obtain the charge injection / redistribution within the metal - molecule - metal junction , the molecular screening of the applied field and the resulting voltage drop , the modification of the molecular states by the applied field , the non - equilibrium occupation of molecular orbitals , the current / conductance - voltage characteristics for the given contact geometry .
the calculation we present is performed using a recently developed self - consistent matrix green s function ( scmgf ) method @xcite which is based on the non - equilibrium generalization of the quasi - particle green s function theory @xcite and uses a finite local - orbital basis set .
by replacing the quasi - particle exchange - correlation self - energy @xcite with the exchange - correlation potential within the density functional theory @xcite , the well - established technique of self - consistent field theory of molecular electronic structure can then be utilized for transport calculations @xcite .
the real - space formulation of our approach allows us to provide an intuitive and coherent physical picture of the molecular tranport by analyzing the device elctronic processes both at the atomic - scale and on the basis of individual molecular orbitals .
we view the molecules as comprised of chemically well - defined atomic groups and interpret their electrical characteristics in terms of the response of these atomic groups to the perturbation induced by the metal - molecule coupling and the applied electric field .
we emphasize the insight obtained with such atomic - scale analysis and show the important effect of atomic - scale charge and potential inhomogeneity on device characteristics at the molecular scale , which may otherwise be obscured by the molecular - level analysis treating the molecules as a whole .
in particular , two important conclusions come out of this work : ( 1 ) the adsorption - induced modification of molecular states is larger than the field - induced effect unless we go to large bias , so accurate modeling of the electronic processes at equilibrium is critical for determining the low - bias transport characteristics ; ( 2 ) the effect of both the metal - molecule coupling and the applied electric field in turning the individual molecular orbitals into effective conduction channels depends on the detailed charge and potential distribution across the metal - molecule - metal junction and may be different for different molecular orbitals .
`` engineering '' the charge and potential inhomogeneity as commonly done in quantum semiconductor heterostructures will be equally important at the molecular scale .
both aspects have been largely neglected in the past .
we therefore hope the results obtained here will provide a useful guide regarding the nature of electron transport at the ultimate limit of device scaling and the prospect of device engineering through molecular design .
the details of the self - consistent matrix green s function theory have been described elsewhere @xcite , so we only give a brief summary here to clarify the assumptions and approximations and to show how the physical observables are computed .
similar but different methods based on the negf approach have been developed independently by other research groups @xcite .
we feel that this approach provides a natural link between quantum transport , first - principles electronic structure theory and qualitative molecular orbital theory @xcite . due to metallic screening in the electrodes ,
the charge and potential perturbations induced by molecular adsorption extend only over a finite region into the metal surface @xcite .
we define an `` extended molecule '' which includes both the molecule and the surface atoms perturbed by molecular adsorption @xcite .
the surface atoms also form an adiabatic reflectionless contact with the rest of the electrodes , which can then modeled as infinite electron reservoirs commonly assumed for mesoscopic transport problems @xcite .
the size of the `` extended molecule '' is chosen such that charge neutrality is maintained approximately under both zero and finite biases .
the central quantities in the negf theory are the retarded and correlation green s function @xmath3 and @xmath4 @xcite .
we expand both the wavefunctions @xmath5 and the green s functions using a finite set of local orbital functions @xmath6 , @xmath7 where @xmath8 is the spin index .
we obtain the retarded and correlation matrix green s function by solving the keldysh - kadanoff - baym ( kkb ) equation in the matrix form @xcite : @xmath9f(e-\mu_{l } ) + i[g^{r;\sigma}(e)\gamma_{r;\sigma}(e)g^{a;\sigma}(e)]f(e-\mu_{r } ) \label{kkb}\end{aligned}\ ] ] where the effect of the contact is incorporated as the self - energy operators @xmath10 , @xmath11 here @xmath12 are the surface green s functions of the left ( l ) and right ( r ) contacts ( the electrodes with the perturbed surface atoms _ _
r__emoved ) which can be obtained from that of the semi - infinite surface . @xmath13
represents the part of the fock matrix contributed by the charge distributions ( both nuclei and electron ) in the `` extended molecule '' only , while @xmath14 represents the long - range coulomb potential due to the equilibrium charge ( ionic and electronic ) distribution in the contact region , which includes the linear voltage drop due to the applied bias .
the @xmath15 are overlap matrices .
the applied potential @xmath16 changes the charge and potential in the `` extended molecule '' , requiring a self - consistent solution .
given the electron density or the density matrix in the `` extended molecule '' , the calculation of the fock matrix @xmath17 is the same as that of standard molecular electronic structure calculation , which greatly facilitates the implementation of our procedure using standard molecular electronic structure codes @xcite .
the self - consistent calculation proceeds by computing the input density matrix to the next iteration from the correlation matrix green s function computed in the current iteration : @xmath18 _ the density matrix is simply the energy integration of the matrix correlation function_. the integration over energy can be performed conveniently in the complex energy plane @xcite .
once the self - consistent calculation converges , we can calculate all the physical observables from the matrix retarded green s function and the density matrix .
we calculate the charge density using @xmath19 from which we can obtain the electrostatic potential in the molecular junction from the poisson equation .
we can also calculate charges associated with each atom using becke s atomic - partition scheme @xcite , @xmath20 here the atomic weight function @xmath21 , which satisfies @xmath22 everywhere in space , is determined such that it is centered on the atom @xmath23 and is non - negligible only in a region close to its atomic center @xcite . the local density of states ( ldos ) gives the energy - resolved charge density distribution : @xmath24 \phi_{i}(\vec r ) \phi_{j}^{*}(\vec r),\ ] ] the spatial integration of ldos gives the density of states , @xmath25 s\},\ ] ] to identify the contribution of the individual molecular orbitals to the total density of states , we project it onto the basis of molecular orbtias .
this is done by transforming @xmath26 into the basis of the molecular orbitals obtained by diagonalizing the ( self - consistent ) fock matrix corresponding to the molecule .
the imaginary part of the diagonal element of the transformed @xmath26 gives the projected density of states ( pdos ) of the corresponding molecular orbitals .
the peak position of the pdos characterizes the perturbed molecular energy level , while the broadness reflects the coupling strength of the molecular oribitals with the metallic surface states .
a critical question in molecular transport is how the electron occupation of the molecular orbitals changes as the molecule is driven out of equilibrium by a finite bias voltages .
the information is contained in the non - equilibrium density matrix @xmath27 ( eq . [ dmne ] ) .
transforming to the basis of molecular orbitals , the diagonal elements of the density matrix give the the non - equilibrium electron occupation of the corresponding molecular levels . a general formula for the current through a mesocopic system with arbitrary interaction in contact with two non - interacting electrodes is : @xmath28 \nonumber \\ & + & [ f(e-\mu_{l})\gamma_{l}^{\sigma}(e , v)-f(e-\mu_{r } ) \gamma_{r}^{\sigma}(e , v ) ] a^{\sigma}(e , v ) ] \}\end{aligned}\ ] ] where @xmath29 is the spectral function .
since the only scattering mechanism in the coherent transport regime is that by the contacts ( introducing additional scattering mechanisms within the molecule will lead to additional terms in self - energy and current ) , we arrive at the familiar landauer - type current formula : @xmath30 \label{iv}\ ] ] where the transmission probability @xmath31 through the molecule @xcite is obtained from : @xmath32= tr[\gamma_{l}^{\sigma}(e , v)g^{r;\sigma}(e , v)\gamma_{r}^{\sigma}(e , v ) [ g^{r;\sigma}]^{\dagger}(e , v)],\ ] ] note that equations ( [ iv]-[tev ] ) hold _ for both orthogonal and non - orthogonal basis functions_. to identify the contribution of individual molecular orbitals to the transmission , we transform the transmission matrix @xmath33 into the the basis of the molecular orbitals whose diagonal elements give the transmission probability through the corresponding molecular orbitals .
combined with the spatially resolved ldos and charge density , the total dos , the transmission coefficient and their projection onto individual molecular orbitals provide a set of useful qualitative analysis tools to establish the connection between the molecular electronic structure and the transport characteristics of the metal - molecule - metal junction , similar to the use of qualitative molecular orbital theory in quantum chemistry @xcite .
a caveat of projecting the physical observable onto the individual molecular orbitals is that phase interference information is lost .
only the summation over all molecular orbitals ( conserved by the matrix transformation ) , not the individual components , corresponds to the quantum mechanical physical observable .
the situation is similar to what we met when we try to decompose the electron density of the molecule into the contribution of individual atoms through a mulliken - type population analysis or other charge partition schemes as we used here . only the total density but not the atom - partitioned density corresponds to a physical observable . within the coherent transport model , the only temperature effect is from the two electrode fermi - dirac distributions .
we can separate the current into two components , the `` tunneling '' component @xmath34 and the `` thermionic emission '' component @xmath35 as follows , @xmath36 de t(e , v)[f(e-\mu_{l})-f(e-\mu_{r})]\ ] ] we choose the electrostatic potential in the middle of the ( empty ) bimetallic junction as the energy reference , then the equilibrium fermi - level @xmath37 is fixed at the negative of the metal work function @xmath38 for single - crystal gold electrodes .
the electrochemical potential of the two electrodes is fixed by the applied bias voltage @xmath39 at @xmath40 and @xmath41
the fermi - level positions are fixed by the bimetallic junction alone without the molecular insertion_. the voltage drop across the molecular junction is determined by the molecular response to the metal - molecule coupling and the applied bias voltages .
the bias polarity is chosen such that for positive bias the electron is injected from the right electrode . in this work
, the electronic structure of the `` extended molecule '' is described using the becke - perdew - wang(bpw91 ) @xcite parameterization of the spin - polarized density - functional theory ( sdft ) @xcite within the generalized - gradient approximation ( gga ) @xcite .
we also replace the atomic core by an appropriate pseudopotential @xcite with the corresponding optimized gaussian basis sets @xcite .
the geometry of the adsorbed molecule is taken to be the same as the singlet geometry of the free molecule optimized at the bpw91/@xmath42 level ( we asuume the bare molecule to be the molecular biradical with the end h atoms removed ) .
the adsorption geometry is chosen such that the end atoms sit in front of the center of the triangular pad of the three gold atoms on the au@xmath1111@xmath2 surface ( the end sulfur - suface distance is @xmath43 ) .
six nearest - neighbor gold atoms on each metal surface are included into the `` extended molecule '' . within the range of second - nearest - neighbor coupling ,
there are @xmath44 metal atoms in the first surface layer and @xmath45 metal atoms on the second surface layers on each side coupled to the `` extended molecule '' .
only the blocks of the surface green s function corresponding to these atoms are needed , which in turn can be calculated from the surface green s function of the semi - infinite metal using tight - binding method parametrized by fitting accurate bulk band structure calculations @xcite .
the results and conclusions given in the following are not affected significantly by small changes in the molecular and adsorption geometry .
the structures of the molecule junction are shown in figs .
( [ xuefig2 - 1 - 1 ] ) and ( [ xuefig2 - 1 - 2 ] ) .
the calculation is performed using a modified version of gaussian98 @xcite .
the problem at equilibrium is a generalization of the familiar chemisorption problem in surface science since two metallic surfaces are involved . for a molecular orbital to be turned into an effective conduction channel of the metal - molecule - metal junction ( within the coherent transport regime ) , it needs to couple well to both electrode states .
its energy must lie within the energy window determined by the bias voltage and the thermal broadening through the electrodes fermi - distribution .
both aspects are affected by the molecular adsorption , which may well be more significant than the subsequent application of the bias voltage , as detailed in subsequent sections .
the transport property of the molecular junction is determined mainly by the hybridization of the surface metal states with the frontier molecular orbitals , i.e. , the molecular states whose energies lie closest to the metal fermi - level .
so we start by illustrating in fig .
( [ xuefig3 - 1 - 1 ] ) and fig .
( [ xuefig3 - 1 - 2 ] ) the charge distribution of the homo-1,homo , lumo and lumo+1 states of the two isolated molecules ( with the end h removed ) . for both molecules
, the homo-1 states are localized on the end sulfur atoms , the homo states are delocalized over entire conjugated backbone while the lumo and the lumo+1 states are benzene based .
the lumo state of bpd also has a finite weight on the end sulfur atoms .
both molecules can be viewed as heterostructures " composed of the end sulfur atoms and the benzene rings . due to the non - coplanar geometry of the two benzenes in bpd molecule ( the torsion angle is @xmath46 ) ,
the orbital overlap between the corresponding @xmath0 electrons is weak .
our analysis of the electrical transport through the two molecules will be based on the electrical response of these chemical units .
the adsoption - induced charge transfer across the metal - molecule interface is the central quantity for the device at equilibrium since the linear - response transport probes the equilibrium charge distribution of the molecular junction .
the perturbation introduced by the metal - molecule coupling is largely a localized interaction .
the corresponding charge - transfer process is determined by the interfacial chemistry and can be understood qualitatively from the local bonding analysis across the metal - molecule interface as in the chemisorption problem @xcite . due to the identical end group and bonding configuration across the metal - molecule interface , both the magnitude and the spatial distribution of the transferred charge for the two molecules are quite similar .
this is illustrated in figs .
( [ xuefig3 - 2 - 1]-[xuefig3 - 2 - 2 ] ) , where we plotted the difference between the self - consistent charge distribution in the gold - molecule - gold junction and the charge distribution in the isolated molecule _ plus _ the charge distribution of the isolated bimetallic contact . for clarity and to aid visualization ,
we have plotted both the regions where charge accumulation and depletion occur and the spatial distribution of the transferred charge as a function of position in the x - y plane ( defined by the left benzene ring ) .
the charge transfer process involves mainly the end sulfur atoms and neighbor carbon atoms and decays rapidly as we move away from the metal - molecule interface .
this is clearly seen for the bpd molecule where the charge perturbation induced by the two metal - molecule contacts decays into the interior of the molecule without effectively interfering with each other ( fig .
[ xuefig3 - 2 - 2 ] ) .
note this is not necessarily due to the non - coplanar geometry .
the decay of charge perturbation is already obvious for the pdt while for bpd it becomes negligible before reaching the inter - benzene bonding region .
the number of electrons increases in the sulfur - gold bonding region due to the rehybridization of the sulfur @xmath47 orbital in forming gold - sulfur bond and the transfer of charge from gold atom to sulfur atom due to electronegativity difference .
electron density also increases in the neighbor carbon @xmath47 orbital since the homo level has large weights on both the sulfur and carbon @xmath47 orbitals .
the electron density decreases in the sulfur - carbon @xmath8-bonding region , the sulfur @xmath48 orbitals and also the gold @xmath49 orbitals .
this is because the formation of surface gold - sulfur bond involves charges originally residing in the sulfur @xmath48 and gold @xmath49 orbitals thus weakening the bonding between sulfur and carbon .
the direction of charge transfer is from the gold electrodes to the molecules .
accompanying the charge transfer across the gold - molecule interface , the electrostatic potential also changes .
the difference of the electrostatic potential in the junction and the electronic potential in the isolated molecule _ plus _ that in the isolated bimetallic contact gives the potential perturbation due to the formation of the contact .
this is plotted in figs .
( [ xuefig3 - 2 - 3 - 1 ] ) and ( [ xuefig3 - 2 - 3 - 2 ] ) . the net transfer of electrons into the molecule increases the electrostatic potential inside the molecule creating a potential barrier between the metal surface and the end sulfur atom .
there is also a potential well between the sulfur atom and the benzene ring because of the decrease of electron density in the sulfur - carbon bonding region .
although the pattern of charge transfer at the metal - molecule interface is similar for the two molecules , the long - range electrostatic potential perturbation is quite different . for the bpd molecule ,
the electrostatic potential profile inside the the two benzene rings becomes quite complicated which creates additional barrier for the electron motion within the molecule core . due to the non - coplanar geometry ,
the barrier for injection from the right metal into the right half of the molecule is larger than that from the left metal to the left half ( the right benzene ring gives a slightly smaller orbital overlap with the right electrode ) .
since the probability of electron tunneling through a potential barrier is sensitive to its shape , the charge - transfer induced potential change will have a profound influence on the electron transmission through the metal - molecule - metal junction .
the presence of the potential barrier at the interface ( and also in the molecule core for bpd ) will affect the charge redistribution within the molecule when an additional field is applied , since it impedes the flow of electrons across the metal - molecule - metal junction .
this will be discussed in the next section .
the coupling between the gold electrodes and the molecule ( reflected in the self - energy operator of the electrodes ) and the induced potential change in the the molecule ( reflected in the self - consistent molecular fock matrix ) modify both the charge distribution and the energy of the molecular states relative to the metal fermi - level .
they also broaden the discrete molecular electronic spectrum into a quasi - continuous one .
these effects can be analyzed through the local density of states , the transmission coefficient and their projection onto the individual molecular orbitals .
the transmission versus energy ( t - e ) and the projected dos ( pdos ) corresponding to the five frontier molecular states of the two molecules ( lumo+1 , lumo , homo , homo-1 and homo-2 ) closest to the fermi - level for spin - up electrons are shown in figs .
( [ xuefig3 - 3 - 1 - 1 ] ) and ( [ xuefig3 - 3 - 1 - 2 ] ) .
the molecules studied here contain an even number of electrons and the gold electrode is non - magnetic , so the spin - up and spin - down channels show identical characteristics .
for other molecules in contact with ferromagnetic electrodes , the spin - resolved analysis is essential for understanding the transport characteristics . from both the transmission versus energy curve and the projected dos curve , we find the fermi - level lies closer to the homo state than to the lumo state for both molecules .
comparing the peak position in the pdos plot with the energy levels of the isolated molecule , we find that the contact with the metallic electrodes significantly shifts the energy levels of the frontier molecular states . the change is found to be larger for the occupied states than for the unoccupied states since the unoccupied states are benzene based ( which also leads to sharper peaks in the pdos plot ) .
the contact with the electrodes also modifies the charge distribution of the molecular states .
this is illustrated by examining the ldos at energies corresponding to the peak positions of the pdos of the homo and lumo for both molecules in figs .
( [ xuefig3 - 3 - 2 - 1 ] ) and ( [ xuefig3 - 3 - 2 - 2 ] ) .
we show both the shape and the spatial distribution of the surface - perturbed molecular states for pdt .
the spatial distribution is obtained by integrating the 3-d ldos with respect to the z - axis and plotted as a function of position in the xy - plane .
the perturbation of the molecular states due to the coupling to the electrodes is different for the two molecules .
in general , the surface - induced change in the charge distribution associated with the molecular states correlates closely with the change in the electrostatic potential . for the pdt molecule ,
the large increase of the electrostatic potential in the middle of the benzene rings pushes the electrons located on the carbon atoms to the peripheral hydrogen atoms for both the homo and the lumo states ( fig .
[ xuefig3 - 3 - 2 - 1 ] ) .
the lumo level remains highly localized giving rise to sharp peak in both pdos and t - e plots . for the bpd molecule
, electrons also move from the right benzene to the left benzene for both the homo and lumo states due to the smaller potential arising from the difference in local geometries .
similar analysis applies to other molecular states .
the peak positions in the pdos plots of the homo and lumo states align closely with the two peak positions in the t - e plot around the fermi - level , corresponding to the onset of resonant transmission . by projecting the total transmission coefficient onto the individual molecular orbitals
, we find the first transmission peak above @xmath50 ( at @xmath51 ) arises mainly from the nearly degenerate lumo and lumo+1 states .
the first transmission peak below @xmath50 ( at @xmath52 ) instead arises mainly from the homo state .
in addition to the transmission peak at the homo and lumo states , there is also a transmission peak in the homo - lumo gap of the pdt molecule at energy @xmath53 ( fig .
[ xuefig3 - 3 - 1 - 1 ] ) .
this peak corresponds to transmission through metal - induced gap states ( migs ) due to the hybridization of metal surface states with the occupied molecular states , which is significant for such a short molecule as pdt .
the projected transmission analysis shows contribution coming mainly from the homo and also other occupied molecular states .
the nature of the migs state is most clearly seen in the corresponding ldos plot ( fig .
[ xuefig3 - 3 - 3 ] ) , which shows similar charge distribution to the homo state .
this is the case for both the ldos and the transmission coefficient throughout the homo - lumo gap region including that at the fermi - level .
for such a short molecule as pdt , the transmission through the metal - induced gap states in the homo - lumo gap is significant .
a similar situatuation occurs also for the bpd molecule except that the transmission through the homo - lumo gap of the bpd molecule is much reduced ( fig .
[ xuefig3 - 3 - 1 - 2 ] ) . the molecule is longer and the orbital overlap with the right electrode states is weaker , so the transmission probability in the gap is much smaller .
the magnitude of the transmission coefficient at the fermi - level determines the zero - bias conductance at low temperature .
we find a conductance of @xmath54 and @xmath55 for the pdt and bpd molecules .
since the length of the molecules are @xmath56 and @xmath57 respectively , the resistance is not proportional to the conjugation length , as expected for tunneling transport .
the current - voltage ( i - v ) and the differential conductance - voltage ( di / dv - v or g - v ) characteristics are calculated for the two molecules using the method described in sec.[s2 ] in the bias range from @xmath58 to @xmath59 and plotted in figs .
[ xuefig4 - 1 - 1 ] and [ xuefig4 - 1 - 2 ] . due to the symmetric device structure , the current - voltage and conductance - voltage characteristics for both molecules are nearly symmetric with respect to bias polarity . in the low bias regime ,
current changes approximately linearly with the bias voltage for both molecules , corresponding to tunneling transport in the homo - lumo gap . both tunneling and thermionic emission contribution to the total current are shown . as expected , for tunneling transport through a large barrier ,
the thermionic emission contribution to the current is negligible even at low bias ( shown in the insets ) . since the variation of the thermionic emission contribution with bias voltage is small ( in the coherent transport regime )
, this further reduces its contribution at high bias .
since this is the only temperature dependence in the coherent transport model , we expect the temperature dependence of the coherent current transport through the two molecules to be weak ( ignoring any disorder effects ) . within the coherent transport model , the thermionic emission and correspondingly the temperature dependence of the current transport
can be important only if the the barrier for electron transmission is small . for comparison ,
we have also plotted the conductance - voltage characteristics obtained using the equilibrium transmission coefficient , .i.e . , replacing @xmath60 in eq .
[ iv ] with @xmath61 .
since the gold surface density of states are approximately symmetric with respect to the fermi - level ( the gold surface band around the fermi - level is due primarily to the @xmath62 electrons ) , the difference between the g - v characteristics thus obtained and the self - consistent g - v characteristics reflects the effect of the bias - induced modification of molecular states . for both molecules , only the low - bias conductance - voltage characteristics ( before reaching the first conductance peak ) can be reasonably well reproduced by the equilibrium transmission characteristics .
the deviation in both the magnitude and the peak position of the conductance becomes significant at large bias , there are also more conductance peaks than would be obtained from the equilibrium transmission characteristics .
so any attempt to predicting the nonlinear transport characteristics from the equilibrium transmission characteristics combined with an assumption about the voltage drop will lead to significant error .
the effect of the bias - induced modification of molecular states is also obvious from looking at the shift of the frontier molecular levels by the applied voltage , as shown in fig .
( [ xuefig4 - 1 - 3 ] ) .
the molecular levels plotted are obtained by diagonalizing the molecular part of the self - consistent fock matrices at each bias voltage .
the molecular levels are nearly constant at low voltages , but begin to shift before reaching the first conductance peak .
the voltage at which the shift occurs corresponds to the voltage where major deviation in the g - v characteristics occurs .
an important question in current transport through molecules is which molecular states are responsible for the observed conductance peak . from the position of the plotted molecular levels
, we expect that for both molecules , the peaks in the conductance are due to the _ occupied molecular _ levels , in disagreement with previous calculation where a jellium model of the electrode is used @xcite .
we believe this is because the jellium model underestimates the metal work function which effectively pushes the homo level down relative to the metal fermi - level @xcite .
the shifts in the individual molecular levels are not identical in either direction or magnitude , so a rigid shift in energy levels as occuring in a planar metal - semiconductor contact does not apply here . instead the shift of the individual molecular states will depend on the detailed charge and potential distributions .
for the two molecules considered here , there is an effective reduction in the homo - lumo gap as bias increases ( fig .
[ xuefig4 - 1 - 3 ] ) .
the charge response of the molecule to the applied electrical field is reflected in both the net charge injection into the molecule and the charge redistribution inside the molecule . for pdt molecule , we find there is only slight additional charge injection into the molecule as we apply the bias voltage . for bpd molecules ,
the net charge injection becomes important only when we go to high bias ( @xmath63 ) .
the charge injection under nonzero bias voltage is determined by the balance of charge injection and extraction at the source - molecule and drain - molecule contacts . since for the pdt molecule
the contact geometries are identical and the gold surface density of states are approximately symmetric with respect to the fermi - levels , little net charge accumulation will be induced by the applied bias voltages . for the bpd molecule , the two rings are non - coplanar , but due to the largely localized nature of the charge injection process , the difference in the two contact becomes non - negligible only at high bias voltages .
in addition to the total net charge in the molecule , we can also calculate the nonequilibrium occupation of the individual orbitals from the nonequilibrium density matrix ( eq .
[ dmne ] ) as a functon of the bias voltage ( fig .
[ xuefig4 - 2 - 1 ] ) .
we find the change in the electron occupation is gradual at the voltages corresponding to the conductance peak , where molecular level moves past the fermi level of the electrodes .
the electron occupation of the molecular orbital is fractional due to the broadening of the molecular level by the coupling to the contact .
although the net charge injection due to applied voltage is negligible , there can be significant charge redistribution within the molecule .
hiscalls for closer investigation into the spatial distribution of the charges and potential distribution within the molecule as a function of applied voltage @xcite .
samples of the spatial distributions of the transferred charge and the electrostatic potential drop for both molecules at bias voltage of @xmath64 are shown in fig .
[ xuefig4 - 2 - 2 ] . similar charge and potential distributions are obtained for _ all _ bias voltages .
the spatial distribution of charge transfer is obtained by integrating the difference in electron density at finite and zero biases along the z - axis and plotted as a function of position in the xy - plane ( defined by the left benzene ring ) .
the potential drop is obtained by evaluating the difference between the electrostatic potential at finite and zero biases , which obeys the boundary condition of approaching @xmath65 ( @xmath66 ) at the left ( right ) electrode .
the molecules exhibit different conductance at different voltages , but the calculated charge and potential distributions show similar patterns and _ do not depend on whether the molecules are in low or high conductance state_. the reason is as follows : the charge and potential response of the molecular junction is determined by the total electron population . but to be in high conductance state , one molecular orbital needs to align with one of the metal fermi - levels .
as it moves away from alignment , the change in the electron population is gradual ( fig .
[ xuefig4 - 2 - 1 ] ) , in contrast with the change in conductance .
given the net charge injected into the molecule , the determination of the electrostatic charge and potential response of the molecule in contact with the two electrodes will be equivalent to that of an isolated _ charged _ molecule under the same boundary conditions of charge and potential . the nature of the charge redistribution can be understood readily by partitioning the molecules into the atomic - groups , i.e. , the benzene rings and the end sulfur atoms .
as bias voltage increases , electrons move from the source - side carbon and sulfur to the drain - side carbon and sulfur .
this is due to fact that the @xmath0 electrons in the carbon @xmath67 orbitals can move freely under the applied electric field .
this flow of @xmath0 electrons is impeded at the molecule - drain contact because the the presence of potential barrier there , which inhibits the charge flow into the drain electrode .
similar considerations apply to the molecule - source contact .
this leads to the creation of two resistivity dipoles at the metal - molecule interface , i.e. , the accumulation of electrons on the injecting side and the depletion of electrons on the extracting side of the potential barrier . for the bpd molecule ,
where the two benzene benzene rings are non - coplanar , the charge response of the two benzene rings show identical behavior and are similar to the benzene ring in the pdt molecule .
the weak orbital overlap between the two benzene rings disrupts the flow of @xmath0 electrons across the inter - benzene bonding region , which creates another potential barrier with corresponding resistivity dipole and partially insulates the two benzene rings from each other .
the spatial variation of the current - induced electron redistribution could also have a significant effect on the structural stability of the molecule under high bias .
as shown in fig .
[ xuefig4 - 2 - 2 ] , applying bias voltage increases the electron density in the sulfur - carbon bond at the source - molecule contact but decreases the electron density in the sulfur - carbon bond at the drain - molecule contact .
according to hellman - feynman theorem @xcite , this would lead to stronger attractive forces and consequently shorter sulfur - carbon bond at the source - molecule contact but weaker forces and longer sulfur - carbon bond at the drain - molecule contact . for the bpd molecule , there is also a shift of electron density in the carbon - carbon bond connecting the two benzene rings which may affect the inter - ring spacing . since the magnitude of the electron redistribution increases with the bias voltage , we expect this bias - induced modification of structural change will have stronger effect at high - bias .
the problem of current - induced conformational change is the molecular analogue of the more familiar electromigration problem in metallic systems @xcite .
although this problem has been treated recently by several groups @xcite , we feel that much needs to be done before a satisfactory theory emerges .
the resistivity dipole is a well known concept in mesoscopic electron tranport @xcite and is common in inhomogenous transport media where their presence in the vicinity of local scattering center helps to overcome the barrier for transport and ensure current continuity .
this can lead to a nonlinear transport effect due to the strong spatial variations in local carrier density and transport field .
such nonlinear transport characteristics induced by the device charge and potential inhomogeneity are a hallmark of band `` engineering '' through mesoscopic semiconductor heterostructures @xcite .
molecular junctions are an analogue of the mesoscopic `` heterostructures '' , since the choice of the component functional and structural groups allows the possibility of `` engineering '' charge and potential inhomogeneity within a single molecule , suggesting the possibility of device `` engineering '' through molecular design and permitting the extension of device concepts to the molecular scale .
this effect can be further elucidated by examining the potential response of the two molecules .
the mobile @xmath0-electrons of the benzene ring effectively screen the applied field , leading to rather flat electrostatic potential drop across the benzene rings .
the electrostatic potential drops rapidly in the molecule - electrode contact region , and becomes flat again as it approaches the boundary of the electrode where strong screening of the applied electric field occurs .
for the pdt molecule , the majority of the voltage therefore drops across the molecule - electrode contact region @xcite .
since the unoccupied molecular states of the pdt molecule are located mainly on the benzene ring where the screened electric field is small , their energy levels do nt change much as bias voltage increases .
the homo-1 state shows stronger voltage dependence than the homo state because it is end - sulfur based where the potential variation is the strongest . for the bpd molecule , although the @xmath0 electrons of each benzene ring can screen the applied field effectively , the potential barrier between the two benzene rings prevents the electron from flowing freely across , leading to significant voltage drop across the carbon framework . as a result ,
there exists strong spatial variation of the electrostatic potential across the molecule core in addition to the metal - molecule contact .
since the four frontier molecular states of the bpd molecule are either sulfur - based or have charge distributed across the entire molecule core , they all show clear voltage dependence . an empirical model of rigid shift of the molecular level relative to the fermi - levels of the two electrodes has often been assumed @xcite , with the proportion of the voltage drop being associated with the contact geometry .
it is clear that this model can only be utilized as a crude check of the device characteristics , since it neglects the possible voltage drop across the molecules and the different effects this may have on different molecular states
. the molecular charge and potential response to the applied electrical field will affect the transport characteristics in two ways : ( 1 ) it shifts the molecular level relative to the fermi - levels of the two electrodes ; ( 2 ) it modifies the charge distribution of the molecular states and therefore their capability for carrying current .
the bias - induced modification of molecular states can be analyzed by examining the local density of states and its projection onto the individual molecular orbitals at finite bias voltages .
a snapshot of the bias dependence of the transmission characteristics and the projected dos corresponding to the five frontier molecular orbitals is shown for the pdt molecule at bias voltages of @xmath68 ( fig .
[ xuefig4 - 3 - 1 ] ) and for the bpd molecule at bias voltages of @xmath69 ( fig . [ xuefig4 - 3 - 3 ] ) to illustrate their voltage dependence .
we have also shown for pdt molecule the ldos at energies corresponding to the peak position in the pdos plots of the homo and lumo states at bias voltage of @xmath70 ( fig .
[ xuefig4 - 3 - 2 ] ) . for the pdt molecule changing from zero bias ( fig .
[ xuefig3 - 3 - 1 - 1 ] ) to @xmath71 ( fig .
[ xuefig4 - 3 - 1 ] ) , both the peak positions in the pdos and the electron distribution associated with the lumo states do nt change much .
the homo energy level also does nt change much , but there is a shift in charge distribution from the right end sulfur atom to the right carbon atom ( not shown here ) . the energy of the homo-1 state increases with bias , leading to a decrease in the energy spacing between the homo and homo-1 states . at @xmath72 ,
the transmission versus energy of the pdt molecules reaches a peak at energy corresponding to the drain fermi - level at @xmath73 , giving rise to the first peak in the conductance .
as we increase the bias voltage further from @xmath71 to @xmath64 ( fig .
[ xuefig4 - 3 - 1 ] ) , the energy levels of the homo and homo-1 states shift gradually toward the equilibrium fermi - level @xmath50 , increasing the transmission coefficient there .
as we further increase the bias to @xmath74 , the charge distribution ( the ldos ) at energies corresponding to the peak position of the lumo state also changes , developing large weight on the end sulfur @xmath75 and the peripheral hydrogen atoms ( fig .
[ xuefig4 - 3 - 2 ] ) . because the homo and homo-1 states move up to the metal fermi - level as the bias increases , the current increases more gradually than would be obtained if the bias - induced modification of molecular states is neglected ( fig .
[ xuefig4 - 1 - 1 ] ) . for the bpd molecule , the first peak in conductance
is reached at @xmath76 ( fig .
[ xuefig4 - 1 - 2 ] ) and fig .
[ xuefig4 - 3 - 3 ] ) .
from zero bias to @xmath77 , the energies of the lumo and homo-1 states decrease and there is a shift of charge from the right benzene to the left benzene ring .
the energy of the homo states increases , and the shift of charge distribution is from the left benzene ring to the right benzene ring ( not shown here ) .
compared to the pdt molecule , the bias - induced modification is stronger despite the longer molecule length , due to the stronger spatial variation of the potential profile .
as we increase the bias to @xmath78 ( fig .
[ xuefig4 - 3 - 3 ] ) , the energy of the lumo continues to decrease , but the energies of both the homo and homo-1 states increase toward the metal fermi - level @xmath50 .
the transmission coefficient at both energies decreases significantly since the electrons are now more localized on the right ring .
a notable change is that a second peak in the pdos of the homo state develops at @xmath79 .
examining the corresponding ldos shows similar charge distribution to that of the homo-2 state with energy of @xmath80 which has charge distribution on both benzene rings ( not shown here ) and correspondingly large transmission probability . increasing the bias to @xmath81 ( fig .
[ xuefig4 - 3 - 3 ] ) further decreases the energy of the lumo state and increase the energies of the homo and homo-1 states . at @xmath74 , the electrons shift from the right benzne ring back to the left for both the homo and homo-1 states ( not shown ) increasing the transmission coefficients there .
since the electrochemical potential of the left ( right ) electrode approaches alignment with the homo-2 and also the lumo state , there is a rapid increase in the conductance as we increases the bias further toward @xmath82 ( fig .
[ xuefig4 - 1 - 2 ] ) .
a common practice in the literature on molecular electronics is to characterize the molecular transport as `` electron transport '' if the conduction is mediated by tunneling through the unoccupied molecuar states , or as `` hole transport '' if the conduction is mediated by tunneling through the occupied molecular states , following the terminology commonly used in semiconductor / organic device and electron transfer research .
here it is important to recognize their differences . for bulk semiconductor devices ,
the concepts of `` electron '' or `` hole '' are associated with introducing shallow - impurity ( dopant ) atoms into the ideal lattice structure , which introduces electron intos the ( unoccupied ) conductand band or removes electrons from the ( filled ) valence band .
the `` electron '' or `` hole '' thus introduced are mobile charge carriers with delocalized wavefunctions .
the system as a whole remains charge neutral , since the charges associated with these carriers are compensated by the impurity ions left behind .
therefore , the type of charge transport is an equilibrium property depending only on the type and amount of dopant atoms introduced .
by contrast , for organic devices , the system is often not doped .
charge transport occurs by injecting charge carriers into the system from the electrodes .
the type of transport depends on the charging state of the molecule and the nonequilibrium injection of charge carriers into the neutral system . for both cases ,
the type of transport is determined by the change in the occupation of electron states through doping or contact injection , indpendent of the nature of charge transport itself ( band transport or polaron hopping ) .
the situation for coherent molecular transport is quite different . for the symmetric molecular devices considered here , both the total number of electrons ( the charging state of the molecule ) and the occupation of the individual molecular orbitals change little with the applied bias .
although a fractional amount of charge is transfered from the gold to the molecule upon electrode contact , the transfered charge is localized in the interfacial region and characterizes the changes in the interfacial bond .
the change in the occupation of molecular states is fractional and gradual , either upon contact to the electrodes at equilibrium or upon application of a nonzero bias voltage out of equilibrium ( fig .
( [ xuefig4 - 2 - 1 ] ) ) .
this is due to the quantum mechanical nature of the coherent tunneling transport , where we can not characterize the tunneling electron as being physically injected into the molecule and subsequently extracted out .
therefore , it is more appropriate to characterize molecular transport through the resonant molecular states without associating it with specific transport types .
the focus of the present series of work is on current transport through a single molecule in contact with two metallic electrodes .
this corresponds closely to the molecular transport measurements using atomic - size break junctions @xcite , where either one or several molecules are probed . however , many molecular transport experiments are performed on molecular monolayers where thousands or tens of thousands of molecules are probed by the contacts to the electrodes @xcite .
such experiments with monolayer configuration present a quite different situation from the single molecule configuration considered here . besides the additional complexity of inter - molecular interactions within the monolayer , the most important difference lies in the interface electrostatics .
note that the same boundary condition for the electrostatic potential across the molecular junction applies to both the single molecule configuration and the monolayer configuration : deep inside the electrodes they approach the bulk value , which can be shifted rigidly with respect to each other by the applied bias voltage .
for the single molecule configuration , the transfer of charge across the metal - molecule interface is confined in a small region , whose contribution to the electrostic potential decays to zero in regions far away from the molecule .
but for the monolayer configuration , the electrostatic potential in regions far away from the molecule is the superposition of contributions from the transfered charge on a large number of molecules . to satisfy the same boundary condition of the electrostatic potential ,
the charge transfer per molecule in a molecular monolayer can be orders of magnitude smaller than that in the single molecule considered here @xcite .
this is the situation often met in organic electronics , where a monolayer of self - assembled molecules is used to modify the work function of the metallic contacts @xcite .
similar problems have also been considered in the electron transport through metal - carbon nanotube interfaces @xcite .
here it is important to recognize the different physics of the metal - molecule interface in single - molecule devices and monolayer devices , since it may have profound effects on the device functionalities achievable using molecular materials .
we have presented a first - principles based microscopic study of current transport through individual molecules .
the real - space formulation allows us to establish a clear connection between the transport characteristics and the molecular electronic structure perturbed by the metal - molecule coupling and the applied electric field . by separating the problem into equilibrium and non - equilibrium situations
, we identify the critical electronic processes for understanding the linear and non - linear transport characteristics . at equilibrium ,
the critical problem is the charge transfer process upon formation of the metal - molecule - metal contact , which modifies the molecular states and determines the energy level lineup relative to the metal fermi - level .
this is mainly an interface - related process and can be controlled by controlling the contact . out of equilibrium ,
the central problem is the molecular charge response and the consequent molecular screening of the applied electric field , which depends on both the molecule core and the nature of the metal - molecule contact and can be understood by viewing the molecules as `` heterostructures '' of chemically well - defined local groups and analyzing their electrical response to the applied electrical field @xcite author to whom correspondence should be addressed .
email : [email protected] a. aviram and m. a. ratner , chem .
* 29 * , 277 ( 1974 ) .
_ molecular electronics : science and technology _ , edited by a. aviram and m. ratner ( new york academy of sciences , new york , 1998 ) .
c. joachim , j. k. gimzewski and a. aviram , nature * 408 * , 541 ( 2000 ) and references thererein .
m. a. reed , c. zhou , c. j. muller , t. p. burgin and j. m. tour , science * 278 * , 252 ( 1997 ) ; mark a. reed , proc .
ieee * 97 * , 652 ( 1999 ) and references therein .
r. m. metzger et al . , j. am . chem . soc . *
119 * , 10455 ( 1997 ) ; r. m. metzger , acc .
32 * , 950 ( 1999 ) .
j. chen , m. a. reed , a. m. rawlett and j. m. tour , science * 286 * , 1550 ( 1999 ) .
z. j. donhauser et al . , science * 292 * , 2303 ( 2001 ) .
c. dekker , phys . today , page 22 , may 1999 .
r. martel et al .
lett . * 73 * , 2447 ( 1998 ) ; h. t. soh , _ ibid . _ * 75 * , 627 ( 1999 ) .
d. porath , a. bezryadin , s. de vries and c. dekker , nature * 403 * , 635 ( 2000 ) .
s. datta , w. tian , s. hong , r. reifenberger , j. j. henderson and c. p. kubiak , phys .
* 79 * , 2530 ( 1997 ) . y. xue , s. datta , w. tian , s. hong , r. reifenberger , j. j. henderson and c. p. kubiak , phys .
rev . b. * 59 * , 7852 ( 1999 ) .
s. hong , r. reifenberger , w. tian , s. datta , j. henderson and c. p. kubiak , superlatt .
microstruc .
* 28 * , 289 ( 2000 ) .
n. b. zhitenov , h. meng and z. bao , phys .
lett . * 88 * , 226801 ( 2001 ) .
y. xue , s. datta and m. a. ratner , j. chem .
* 115 * , 4292 ( 2001 ) .
y. xue , s. datta and m. a. ratner , chem .
* 281 * , 151 ( 2002 ) .
( lanl archive : xxx.lanl.gov/cond-mat/0112136 ) .
p. s. damle , a. w. ghosh and s. datta , phys/ rev .
b. * 64 * , 201403 ( 2001 ) .
m. di ventra , s. t. pantelides and n. d. lang , phys .
lett . * 84 * , 979 ( 2000 ) ; n. d. lang and ph .
avouris , phys .
b * 64 * , 125323 ( 2001 ) m. di ventra and n. d. lang , phys .
b * 64 * , 45402 ( 2001 ) .
j. taylor , h. guo and j. wang , phys .
b * 63 * , 245407 ( 2001 ) .
m. brandbyge et al . , phys .
b * 65 * , 165401 ( 2002 ) .
a. nitzan , annu .
* 52 * , 681 ( 2001 ) .
e. g. emberly and g. kirczenow , phys .
b * 60 * , 6028 .
l. e. hall et al . ,
j. chem . phys . *
112 * , 1510 ( 2000 ) .
j. m. seminario , a. g. zacarias and j. m. tour , j. am .
soc . * 121 * , 411 ( 1999 ) .
b. c. stipe , m. a. rezaei and w. ho , science * 280 * , 1732 ( 1998 ) ; phys .
lett . * 82 * , 1724 ( 1999 ) .
z. g. yu , d. l. smith , a. saxena and a. r. bishop , phys .
b * 59 * , 16001 ( 1999 ) .
h. ness , s. a. shevlin and a. j. fisher , phys .
b * 63 * , 125422 ( 2001 ) .
l. v. keldysh , sov .
jetp * 20 * , 1018 ( 1965 ) .
d. c. langreth , in _ linear and non - linear electron transport in solids _ , edited by j. t. devreese and e. van doren ( plenum press , new york , 1976 ) .
p. danielewicz , ann . phys . * 152 * , 239 ( 1984 ) . g. d. mahan , phys
. rep .
* 145 * , 251 ( 1987 ) .
h. haug and a - p .
jauho , _ quantum kinetics in transport and optics of semiconductors _ ( springer - verlag , berlin , 1996 ) .
w. g. aulbur , l. jnsson and j. w. wilkins , solid state phys . *
54 * , 1 ( 1999 ) .
_ theory of the inhomogeneous electron gas _ , edited by s. lundqvist and n. h. march ( plenum press , new york , 1983 ) .
r. m. dreizler and e. k. u. gross , _ density functional theory : an approach to the quantum many - body problem _ ( springer - verlag , berlin , 1990 ) .
r. hoffmann , rev .
* 60 * , 601 ( 1988 ) .
t. a. albright , j. k. burdett and m .- h .
whangbo , _ orbital interactions in chemistry _ ( john wiley and sons , new york , 1985 ) .
a. r. williams , p. j. feibelman and n. d. lang , phys .
b * 26 * , 5433 ( 1982 ) .
brivio and m. i. trioni , rev .
phys . *
71 * , 231 ( 1999 ) .
s. n. yaliraki , a. e. roitberg , c. gonzalez and m. a. ratner , j. chem .
111 * , 6997 ( 1999 ) .
m. magoga and c. joachim , phys .
b * 56 * , 4722 ( 1997 ) .
r. zeller , j. deuta and p. h. dederichs , solid state commun .
* 44 * , 993 ( 1982 ) .
a. d. becke , j. chem .
phys . * 88 * , 2547 ( 1988 ) .
y. meir and n. s. wingreen , phys .
lett . * 68 * , 2512 ( 1992 ) .
s. datta , _ electron transport in mesoscopic systems _ , ( cambridge university press , cambridge , 1995 ) j. c. cuevas , a. levy yeyati , and a. martin - rodero , phys .
lett . * 80 * , 1066 ( 1998 ) .
a. d. becke , phys .
a. * 38 * , 3098 ( 1988 ) .
j. p. perdew and y. wang , phys .
rev . b. * 38 * , 12228 ( 1988 ) ; m. ernzerhof , j. p. perdew and k. burke , in _ density functional theory i _ , edited by r. f. nalewajski ( springer , berlin , 1996 ) .
o. gunnarson and b. i. lundqvist , phys .
b * 13 * , 4274 ( 1976 ) ; r. o. jones and o. gunnarson , rev . mod .
phys . *
61 * , 689 ( 1989 ) .
l. r. kahn , p. bybutt and d. g. truhlar , j. chem .
phys . * 65 * , 3826 ( 1976 ) . for carbon and sulfur
, we use the pseudopotential and the corresponding polarized split valence basis sets of w. j. stevens , h. basch and m. krauss , j. chem .
81 * , 6026 ( 1984 ) ; for gold , we use the pseudopotential and the valence basis sets of p. j. hay and w. r. wadt , j. chem .
phys . * 82 * , 270 ( 1985 ) . for the gold atoms lying on top of the semi - infinite substrate ( e.g. describing the atomic - scale structure of the surface ) ,
we use the standard double - zeta basis set as described by p. j. hay and w. r. wadt ( j. chem . phys . * 82 * , 270 ( 1985 ) ) ; for the gold atoms on the surface of the semi - infinite substrate , we follow the standard convention in surface and solid state electronic structure calculation using molecular gaussian - type orbital basis sets to construct a minimum valence basis set by removing the most diffuse gaussian primitive .
a minimum - valence basis set may also be formed including the most diffuse component ( h. basch , unpublished ) , however we find this leads to convergence problem and overcharging of the `` extended molecule '' with charge overflow onto the substrate atoms without affecting the molecule significantly . for further information rearding
the basis set usage for surface and bulk calculations , see the seris of work related to the crystal98 package ( v. saunders , r. dovesi , c. roetti , m. causa , n. harrison , r. orlando , and c. m. zicovich - wilson , universit di torino , torino , 1998 ) , e. g. , y. noel et al . ,
b * 65 * , 14111 ( 2001 ) ; a. kokaly and m. causa , j. phys . : condens .
matter * 11 * , 7463 ( 1999 ) .
d. a. papaconstantopoulos , _ handbook of the band structure of elemental solids _ ( plenum press , new york , 1986 ) .
gaussian 98 , revision a.7 , m. j. frisch , g. w. trucks , h. b. schlegel , g. e. scuseria , m. a. robb , j. r. cheeseman , v. g. zakrzewski , j. a. montgomery jr . , r. e. stratmann , j. c. burant , s. dapprich , j. m. millam , a. d. daniels , k. n. kudin , m. c. strain , o. farkas , j. tomasi , v. barone , m. cossi , r. cammi , b. mennucci , c. pomelli , c. adamo , s. clifford , j. ochterski , g. a. petersson , p. y. ayala , q. cui , k. morokuma , d. k. malick , a. d. rabuck , k. raghavachari , j. b. foresman , j. cioslowski , j. v. ortiz , a. g. baboul , b. b. stefanov , g. liu , a. liashenko , p. piskorz , i. komaromi , r. gomperts , r. l. martin , d. j. fox , t. keith , m. a. al - laham , c. y. peng , a. nanayakkara , c. gonzalez , m. challacombe , p. m. w. gill , b. johnson , w. chen , m. w. wong , j. l. andres , c. gonzalez , m. head - gordon , e. s. replogle and j. a. pople " , gaussian , inc . ,
pittsburgh , pa , 1998 . v. heine , phys .
a*138 * , 1689 ( 1965 ) .
j. tersoff , phys .
lett . * 52 * , 465 ( 1984 ) ; j. tersoff , phys . rev .
b * 30 * , 4874 ( 1984 ) .
w. mnch , rep .
phys . * 53 * , 221 ( 1990 ) .
e. h. rhoderick and r. h. williams , _ metal - semiconductor contact _ , 2nd edition ( oxford university press , oxford , 1988 ) .
i. h. campbell and d. l. smith , solid state phys . * 55*,1 ( 2001 ) .
p. k. h. ho , m. granstrm , r. h. friend and n. c. greenham , adv . mat . * 10 * , 769 ( 1998 ) .
w. li et al . , adv . mat . * 11 * , 730 ( 1999 ) .
v. mujica , a. e. roitberg and m. a. ratner , j. chem .
112 * , 6834 ( 2000 ) .
r. landauer , ibm j. res .
develop .
* 1 * , 223 ( 1957 ) ; * 32 * , 306 ( 1988 ) .
a. wacker , phys . rep .
* 357 * , 1 ( 2002 ) . _
" the force concept in chemistry _ , edited by b. m. deb ( van norstrand reinhold company , new york , 1981 ) .
a. h. verbruggen , ibm j. res .
develop . *
32 * , 93 ( 1988 ) .
f. kassubek , c. a. stafford and h. grabert , phys .
b * 59 * , 7560 ( 1999 ) .
j. reichert et al . ,
bf 88 , 176804 ( 2002 ) .
j. g. kushmerick et al . ,
bf 89 , 86802 ( 2002 ) .
j. park et al . ,
nature * 417 * , 722 ( 2002 ) . w. liang et al . ,
nature * 417 * , 725 ( 2002 ) .
cui et al . , science * 294 * , 571 ( 2001 ) .
a simple estimate can be given in the limit of an ideal ( infinite ) two - dimensional molecular monolayer .
we assume the molecule - metal distance is @xmath83 and each molecule occupies a surface area of @xmath84 on average .
replacing the charge distribution of the two - dimensional monolayer by an uniform charge sheet , a charge transfer of @xmath85 per molecule is enough to give an electrostatic potential drop of @xmath86 across the metal - monolayer junction .
d. janes , private communication . f. lonard and j. tersoff , phys .
lett . * 83 * , 5174 ( 1999 ) .
y. xue and m.a .
ratner , to be published . since our emphasis in this work is in conceptual issues , we have not tried testing or optimizing the more technical aspects of the modeling work such as the choice of the exchange - correlation functionals and the optimum molecular basis sets . for example , the calculations reported in this work were performed using a quite restricted basis set for the metal atoms on the surface of the semi - infinite substrate .
the metal atoms belonging to atomic - scale structures on top of the substrate and the molecule itself are treated using much better basis sets as in standard molecular calculations .
it is clear that the description of the metallic substrate is inadequate . and
obtaining a better balance in the basis set decription between the substrate atoms and the molecule is a difficult and important issue for further improving the accuracy of the modeling of molecular - scale devices within the negf formalism .
such refinement may be needed for obtaining quantitative agreement with specific transport measurement .
in particular , the calculated charger transfer from the gold surface to sulfur may be different depending on the basis sets ( plane waves or extended gaussian basis sets ) as well as the surface models ( semi - infinite crystal , cluster or supercell geometry ) chosen . | we present the first in a series of microscopic studies of electrical transport through individual molecules with metallic contacts .
we view the molecules as `` heterostructures '' composed of chemically well - defined atomic groups , and analyze the device characteristics in terms of the charge and potential response of these atomic - groups to the perturbation induced by the metal - molecule coupling and the applied electrical field , which are modeled using a first - principles based self - consistent matrix green s function ( scmgf ) method . as the first example , we examine the devices formed by attaching two benzene - based molecular radicals phenyl dithiol ( pdt ) and biphenyl dithiol ( bpd)symmetrically onto two semi - infinite gold electrodes through the end sulfur atoms .
we find that both molecules acquire a fractional number of electrons with similar magnitude and spatial distribution upon contact with the electrodes .
the charge transfer creates a potential barrier at the metal - molecule interface that modifies significantly the frontier molecular states depending on the corresponding electron density distribution . for both molecules ,
the metal fermi - level is found to lie closer to the highest - occupied - molecular - orbital ( homo ) than to the lowest - unoccupied - molecular - orbital ( lumo ) .
transmission in the homo - lumo gap for both molecules is due to the metal - induced gap states arising from the hybridization of the metal surface states with the occupied molecular states . applying a finite bias voltage leads to only minor net charge injection due to the symmetric device structure assumed in this work .
but as current flows , the electrons within the molecular junction redistribute substantially , with resistivity dipoles developing in the vicinity of potential barriers . only the delocalized @xmath0-electrons in the benzene ring
can effectively screen the applied electric field .
for the pdt molecule , the majority of the bias voltage drops at the metal - molecule interface .
but for the bpd molecule , a significant amount of the voltage also drops in the molecule core .
the field - induced modification of the molecular states ( the static stark effect ) becomes significant as the bias voltage increases beyond the linear - transport region .
a bias - induced reduction of the homo - lumo gap is observed for both molecules at large bias .
the stark effect is found to be stronger for the bpd molecule than the pdt molecule despite the longer length of the former .
for both molecules , the peaks in the conductance are due to electron transmission through the occupied rather than the unoccupied molecular states .
the calculation is done at room temperature , and we find the temperature dependence of the current - voltage characteristics of both molecules is negligible . |
in physics the search and study of exactly solvable systems have a long history . in non relativistic quantum mechanics
a problem thoroughly investigated is to find potentials for which we can solve exactly their schrdinger equations in terms of special functions , since they can be used as a basis in the analysis of more realistic problems and their solutions can be exploited to test numerical calculations .
several methods have been developed to obtain exact solutions to the schrdinger equation @xcite@xcite , here we mention the methods based on the point canonical transformations @xcite , on the darboux transformations @xcite , @xcite , and on the supersymmetric quantum mechanics @xcite@xcite ( see also the factorization method @xcite@xcite related to the supersymmetric quantum mechanics ) . at present time we know many potentials for which we can solve exactly their schrdinger equations @xcite@xcite , @xcite@xcite .
furthermore , there are potentials for which we can find their solutions in terms of special functions only if the parameters satisfy some condition @xcite@xcite .
they are called conditionally exactly solvable potentials ( ces potentials ) . in this work ,
our main objective is to give two potentials defined on the half line @xmath0 such that each linearly independent solution of the schrdinger equations includes a sum of two confluent hypergeometric functions .
as far as we know these potentials are not previously studied @xcite@xcite , @xcite@xcite . here
we notice that the method used in this paper works simultaneously with the two schrdinger equations of the partner potentials .
the potentials that we study in this paper are algebraic modifications of the coulomb potential ( see the expression ( [ e : ces potentials ] ) )
. these potentials may be useful in the study of physical systems for which we expect deviations from the coulomb potential near @xmath1 , but that are of coulomb type as @xmath2 .
it is convenient to notice that potentials with negative fractional powers have been used as models to study the quark - antiquark interaction @xcite and sharp resonances @xcite . from the shape of the potentials ( see figure 2 below ) we think that these can be useful models to study scattering and tunneling problems @xcite . furthermore , in an interval the shape of the potential that we call @xmath3 reminds us to the effective potentials that appear in the analysis of the propagation of the dirac field in curved spacetimes , for example , in the schwarzschild black hole @xcite .
we organize this paper as follows . in sect .
[ s : method ] we expound the potentials that we study through this work and we solve the schrdinger equations of these potentials .
we also find some properties of these solutions .
finally in sect .
[ s : discussion ] we discuss other properties of these potentials .
here we show that for @xmath4 and for the potentials @xmath5 we can solve the schrdinger equations @xmath6 in terms of confluent hypergeometric functions . as @xmath7 . ] for the potentials ( [ e : ces potentials ] ) the superpotential @xmath8 is equal to @xcite@xcite @xmath9 thus , in the standard language of the supersymmetric quantum mechanics , @xmath10 and @xmath3 are partner potentials @xcite@xcite . the partner potentials ( [ e : ces potentials ] )
are ces since the constants @xmath11 and @xmath12 that multiply to the factors @xmath13 and @xmath14 satisfy the condition @xmath15 as far as we see the partner potentials ( [ e : ces potentials ] ) do not appear in refs .
@xcite@xcite , @xcite@xcite that enumerate the previously known exactly solvable potentials of the schrdinger equation .
moreover the ces potentials ( [ e : ces potentials ] ) remind us to those of refs .
@xcite , @xcite @xmath16 where @xmath17 , @xmath18 , @xmath19 , @xmath20 are constants , since they include negative fractional powers of the variable @xmath21 in addition to negative integral powers of the same variable .
[ figure1 ] for @xmath22 ( solid line ) and for @xmath23 ( broken line ) . ] [ figure2 ] ( solid line ) and the potential @xmath3 ( broken line ) for @xmath24 . ] in what follows we assume that @xmath25 . we notice that near @xmath26 and as @xmath27 the superpotential ( [ e : ces superpotential ] ) and the ces potentials ( [ e : ces potentials ] ) behave as @xmath28 where @xmath29 ( @xmath30 ) means that the quantity goes to zero taking positive ( negative ) values .
we also point out that the ces potentials @xmath31 go to zero in the form @xmath13 as @xmath27 , and they diverge as @xmath14 as @xmath32 .
the potential @xmath10 is strictly positive , whereas the potential @xmath3 crosses the @xmath21 axis at @xmath33 .
since the derivatives of the partner potentials ( [ e : ces potentials ] ) are equal to @xmath34 the potential @xmath10 is a decreasing function , and we get that the derivative of @xmath3 is equal to zero at @xmath35 . this critical point is a maximum .
notice that @xmath36 , that is , the coordinate of the maximum is larger than the point where the potential @xmath3 crosses the @xmath21 axis . for @xmath4 the superpotential @xmath8 is strictly negative and it is an increasing function . for @xmath8 and @xmath31
we illustrate these facts in figs . 1 , 2 . for @xmath37
a straightforward calculation shows that the schrdinger equations ( [ e : schrodinger equation ] ) for partner potentials transform into @xmath38 from these equations we get that the functions @xmath39 must be solutions to the coupled system of first order differential equations @xmath40 and defining the functions @xmath41 and @xmath42 by @xmath43 we find that the schrdinger equations ( [ e : schrodinger equation ] ) of partner potentials simplify to the system of coupled differential equations for the functions @xmath41 and @xmath42 @xmath44 thus to solve the schrdinger equations of the partner potentials ( [ e : ces potentials ] ) we make the change of variable @xmath45 to obtain that eqs .
( [ e : coupled tilde r ] ) take the form @xmath46 from this coupled system we get that the functions @xmath41 and @xmath42 satisfy the decoupled equations @xmath47
\tilde{r}_j = 0,\ ] ] where @xmath48 and @xmath49 ( @xmath50 ) for @xmath41 ( @xmath42 ) . to find the solutions to the previous equations we make the ansatz @xmath51 to get that the functions @xmath52 must be solutions to @xmath53 we notice that the previous equations are of confluent hypergeometric type @xcite@xcite @xmath54 with the quantities @xmath55 and @xmath56 equal to @xmath57 since the parameters @xmath56 are not integral numbers , the functions @xmath58 take the form @xcite@xcite @xmath59 where @xmath60 denotes the confluent hypergeometric function and @xmath61 , @xmath62 are constants . from the values ( [ e : values parameters a b ] ) it is convenient to note that @xmath63 and therefore making the change of variable @xmath64 we can transform eqs .
( [ e : r radial y coordinate ] ) into the hermite differential equations @xcite@xcite @xmath65 with @xmath66 . thus using the linearly independent solutions of eqs .
( [ e : hermite form ] ) we can expand the functions @xmath52 , but to write the solutions of the differential equations ( [ e : r radial y coordinate ] ) we prefer to use confluent hypergeometric functions .
furthermore , to increase the readability , in the expressions that follow we replace the quantities @xmath67 and @xmath68 by their numerical values , but we preserve the parameters @xmath69 and @xmath70 .
we point out that the four constants @xmath61 and @xmath62 are not independent since eqs .
( [ e : coupled tilde r variable y ] ) impose the following restrictions : \a ) if we choose the function @xmath41 as @xmath71 then the function @xmath42 must take the form @xmath72 that is , the constants @xmath73 and @xmath74 fulfill @xmath75 \b ) when we select the function @xmath76 as @xmath77 then the function @xmath42 must be of the form @xmath78 hence the constants @xmath79 and @xmath80 satisfy @xmath81 from these results and considering the definition ( [ e : r tilde definition ] ) we get that the linearly independent solutions @xmath39 to the schrdinger equations of the ces potentials ( [ e : ces potentials ] ) are equal to @xmath82,\end{aligned}\ ] ] and @xmath83 .\end{aligned}\ ] ] using that for the linearly independent solutions of the confluent hypergeometric equation ( [ e : confluent hypergeometric equation ] ) its wronskian is equal to @xcite , @xcite @xmath84 = ( 1-b ) z^{-b } \textrm{e}^z,\ ] ] we find that the wronskian of the solutions ( [ e : zeta one ] ) and ( [ e : zeta two ] ) is ( for @xmath85 ) @xmath86 = \mp \frac{\omega \ , i^{3/2}}{m } ( 2 \omega)^{1/2 } , \ ] ] which is a constant , as expected .
moreover considering eqs .
( [ e : r tilde definition ] ) , ( [ e : coupled tilde r ] ) , and ( [ e : r tilde differential equations ] ) we obtain that the functions @xmath87 are solutions of the equations @xmath88 in a straightforward way we can verify that these are the schrdinger equations of the partner potentials @xmath31 written using the variable @xmath89 defined in the formula ( [ e : y definition ] ) , and hence we find that the functions @xmath58 yield the solutions to the schrdinger equations of the partner potentials ( [ e : ces potentials ] ) .
we have not been able to transform the linearly independent solutions ( [ e : zeta one ] ) , ( [ e : zeta two ] ) and write them in terms of one confluent hypergeometric function , but we think that this fact must be considered in detail . thus we find in explicit form the partner potentials ( [ e : ces potentials ] ) whose linearly independent solutions include the sum of two confluent hypergeometric functions ( see ( [ e : r tilde definition ] ) , ( [ e : zeta one ] ) , and ( [ e : zeta two ] ) ) . in ref .
@xcite some potentials with this property are studied , but they are given in implicit form .
hence the partner potentials ( [ e : ces potentials ] ) expand the previously known examples of potentials whose linearly independent solutions include a sum of ( confluent ) hypergeometric type functions @xcite
. furthermore the ces potentials @xmath31 are multiplicative shape invariant @xcite , @xcite , @xcite . from their expressions ( [ e : ces potentials ] ) we find that they satisfy @xmath90 .
when we compare with the definition of shape invariance for the partner potentials @xcite @xmath91 where @xmath92 , @xmath93 are parameters and @xmath94 is a function of @xmath92 , we find that the partner potentials @xmath31 fulfill @xmath95 , @xmath96 , @xmath97 , and hence , @xmath98 . we notice that the previously published multiplicative shape invariant potentials only are known in series form @xcite , @xcite , in contrast we find the potentials ( [ e : ces potentials ] ) in closed form . in ref .
@xcite is studied the concept of shape invariance with reflection transformations . for the analyzed potentials they consider transformations that include reflections of the coordinates and translations of the parameters .
the relationship @xmath98 between the parameters of the partner potentials ( [ e : ces potentials ] ) reminds us a reflection , but this reflection is on the parameters of the potentials and not on the coordinates , as considered in ref .
@xcite . owing to the ces potentials ( [ e : ces potentials ] )
go to zero as @xmath99 , in this limit @xmath39 behave as @xmath100 where @xmath101 denote the phase shifts of @xmath31 ( @xmath102 are constants ) . using the known relationship between the phase shifts of the partner potentials
@xcite , @xcite @xmath103 where @xmath104 is the limit of the superpotential @xmath8 as @xmath99 ( see ( [ e : limits potentials superpotential ] ) ) , we get that the phase shifts of the ces potentials ( [ e : ces potentials ] ) fulfill @xmath105 where @xmath106 is an integer . from our previous results
we think that a research problem is the search , in explicit form , of potentials for which each linearly independent solution of their schrdinger equations includes a sum of three ( or four , five , ) ( confluent ) hypergeometric functions .
for the case of three ( confluent ) hypergeometric functions we expect that the first linearly independent solution takes the form @xmath107 where @xmath108 are functions of @xmath21 and @xmath109 denotes the ( confluent ) hypergeometric functions ( a similar expression is valid for the second linearly independent solution ) . ) and ( [ e : zeta two ] ) for linearly independent solutions of the schrdinger equation that include two confluent hypergeometric functions . ]
notice that in ref .
@xcite are studied some potentials with this property , but they are given in implicit form .
if we find these potentials , they can be used as a basis to generate new exactly solvable potentials by means of first or higher order supersymmetric quantum mechanics @xcite@xcite .
we thank the support by conacyt mxico , sni mxico , edi - ipn , cofaa - ipn , and research projects ipn sip-20150707 and ipn sip-20151031 . also , we thank the useful suggestions of the anonymous referees . | we give two conditionally exactly solvable inverse power law potentials whose linearly independent solutions include a sum of two confluent hypergeometric functions . we notice that they are partner potentials and multiplicative shape invariant . the method used to find the solutions works with the two schrodinger equations of the partner potentials .
furthermore we study some of the properties of these potentials .
keywords : conditionally exactly solvable potentials ; shape invariance ; confluent hypergeometric equation .
pacs : 03.65.ge , 03.65.nk , 03.65.ca , 02.90.+p |
we consider the system of generalized burgers equations for @xmath0 unknown variables @xmath2 , @xmath3 where @xmath4 is a constant vector in @xmath5 and @xmath6 is the usual inner product in @xmath5 .
we study the solution of ( 1.1 ) with initial conditions @xmath7 when @xmath8 , ( 1.1 ) is a system of nonlinear parabolic equation describing the interplay between nonlinearity and diffusion , @xmath9 being the viscosity parameter . when @xmath10 , the system ( 1.1 ) is hyperbolic with coinciding wave speeds @xmath11 , and the nonlinearity and the nonconservative form makes the initial value problem complex .
indeed the system being nonlinear , solution can not be continued as a smooth solution even when the initial data is smooth .
further if @xmath12 , the product @xmath13 is nonconservative and does not make sense in the usual distributional sense .
the solution should be understood in a generalized sense .
for the case @xmath14 , joseph @xcite used a generalized hopf
cole transformation to linearize the system of equations @xmath15 and solve ( explicitly ) with the conditions @xmath16 in terms of a family of probability measures @xmath17 .
these measures depend on the initial data ( 1.2 ) in a nonlinear and nonlocal manner and takes the form @xmath18}\d y}{\int_{-\infty}^{\infty } \e^{-\frac{1}{\nu } \left[i(y ) + \frac{(x - y)^2}{2t}\right]}\d y},\ ] ] where @xmath19 when @xmath14 , the solution of ( 1.1 ) and ( 1.2 ) was shown to be @xmath20 let @xmath21 be the solution of ( 1.1 ) and ( 1.2 ) given by ( 1.5 ) .
it was shown in @xcite that when @xmath22 is lipschitz continuous , for each @xmath23 , except for a countable @xmath24 the limits @xmath25 exist and is given by the formula @xmath26 where @xmath27 is a minimizer of @xmath28 and @xmath29 is given by ( 1.4 ) .
using some ideas from the earlier works of joseph @xcite and lefloch @xcite , @xmath30 was shown to be a solution of an inviscid case @xmath31 in ( 1.1 ) with initial data from ( 1.2 ) , the nonconservative product was justified in the sense of volpert product @xcite ( see dal maso , lefloch and murat @xcite for a generalization of volpert product ) . in @xcite solution for general initial data is constructed in the sense of colombeau @xcite . the aim of the present note is to study the asymptotic behaviour of the solution for the parabolic ( viscous ) case as well as the hyperbolic ( inviscid ) case .
study of asymptotic behaviour of solutions is important , see @xcite and the references therein for the parabolic case and @xcite for the inviscid case .
when @xmath1 , this system is the celebrated burgers equation and explicit solution and its asymptotic behavior as @xmath32 tends to infinity and diffusion parameter @xmath33 was studied by hopf @xcite .
we show that hopf s analysis give asymptotic form of the solution for the viscous case .
when @xmath34 and @xmath35 , it is well - known from the work of lax @xcite that for solution of inviscid burgers equation with initial data supported in the compact interval @xmath36 $ ] , @xmath37 the solution decays at the rate @xmath38 and support spreads at a rate @xmath39 for large time . from an explicit computation
we will show that the decay rate is not true in general for the present case , but still the support spread at the same rate .
we start with the viscous case .
in this section we study the asymptotic behavior of solution of ( 1.1 ) and ( 1.2 ) when @xmath14 and fixed . on the initial conditions @xmath40 ,
assume that @xmath41 , @xmath42 , @xmath43 , @xmath44 exists and is finite . with these assumptions
, we shall prove the following result .
the solution @xmath45 of ( 1.1 ) and ( 1.2 ) has the following asymptotic behaviour as @xmath32 tends to infinity @xmath46 first we note that the solution @xmath47 of ( 1.1 ) and ( 1.2 ) is given by ( 1.5 ) where the measure @xmath17 is given by eqs ( 1.3 ) and ( 1.4 ) .
writing explicitly the formula , we get @xmath48}\d y}{\int_{-\infty}^{\infty } \e^{-\frac{1}{\nu } \left[i(y ) + \frac{(x - y)^2}{2t}\right]}\d y}.\ ] ] setting @xmath49 , and then making a change of variable @xmath50 and renaming @xmath51 as @xmath52 , we get @xmath53 } \d y}{\int_{-\infty}^{\infty } \e^{-\left[\frac{i(\sqrt{\nu t}(\xi - y))}{\nu } + y^2/2\right ] } \d y}.\ ] ] now we split the integral in ( 2.2 ) in the following manner : @xmath54 } \d y\nonumber\\[.5pc ] & \quad = \int_{-\infty}^{\xi -\delta } u_{0j}(\sqrt{\nu t}(\xi - y))\e^{-\left[\frac{i(\sqrt{\nu t}(\xi - y))}{\nu}+y^2/2\right ] } \d y\nonumber\\[.5pc ] & \qquad + \int_{\xi + \delta}^{\infty } u_{0j}(\sqrt{\nu t}(\xi - y))\e^{-\left[\frac{i(\sqrt{\nu t}(\xi - y))}{\nu}+y^2/2\right ] } \d y\nonumber\\[.5pc ] & \qquad + \int_{\xi-\delta}^{\xi+\delta } u_{0j}(\sqrt{\nu t}(\xi - y ) ) \e^{-\left[\frac{i(\sqrt{\nu t}(\xi - y))}{\nu}+y^2/2\right ] } \d y.\end{aligned}\ ] ] now we fix @xmath55 and study each of these integrals as @xmath32 tends to infinity .
we have under the assumptions of the theorem on @xmath29 and @xmath56 , as @xmath32 tends to infinity : @xmath57 } \d y \approx \e^{-\frac{i(+\infty)}{\nu}}u_{0j}(\infty)\int_{-\infty}^{\xi -\delta } \e^{-y^2/2 } \d y.\\[.5pc ] \hskip -3pc & \int_{\xi + \delta}^{\infty } u_{0j}(\sqrt{\nu t}(\xi - y ) ) \e^{-\left[\frac{i(\sqrt{\nu t}(\xi - y))}{\nu } + y^2/2\right ] } \d y \approx \e^{-\frac{i(-\infty)}{\nu}}u_{0j}(-\infty)\int_{\xi + \delta}^{\infty } \e^{-y^2/2 } \d y , \\[.5pc ] \hskip -3pc & \limsup_{t\rightarrow \infty } \left\vert\int_{\xi - \delta}^{\xi + \delta } u_{0j}(\sqrt{\nu t}(\xi - y ) ) \e^{-\left[\frac{i(\sqrt{\nu t}(\xi - y))}{\nu } + y^2/2\right ] } \d y\right\vert = o(\delta).\end{aligned}\ ] ] now let @xmath32 tend to infinity and then @xmath58 tend to @xmath59 in ( 2.3 ) .
we get @xmath60 } \d y & \approx \e^{-\frac{i(+\infty)}{\nu}}u_{0j}(\infty)\int_{-\infty}^{\xi } \e^{-y^2/2 } \d y\\[.5pc ] & \quad + \e^{-\frac{i(-\infty)}{\nu}}u_{0j}(-\infty)\int_{\xi}^{\infty } \e^{-y^2/2 } \d y. \end{split}\end{aligned}\ ] ] similarly @xmath61 } \d y \approx \e^{\frac{-i(+\infty)}{\nu } } \int_{-\infty}^{\xi } \e^{-y^2/2 } \d y + \e^{\frac{-i(-\infty)}{\nu } } \int_{\xi}^{\infty } \e^{-y^2/2 } \d y.\ ] ] we observe that due to our assumption on @xmath29 , this limit in ( 2.5 ) is a postive real number and hence letting @xmath32 tend to infinity in ( 2.2 ) and using ( 2.4 ) and ( 2.5 ) we get the result ( 2.1 ) .
the proof of the theorem is complete .
@xmath62 an interesting case here is when the initial data @xmath63 satisfies the following conditions .
@xmath64 , @xmath65 are nonzero and there is a cancellation in @xmath66 so that this quantity is integrable .
in this section we study the asymptotic behaviour of vanishing viscosity solutions of @xmath67 with initial data @xmath68 for @xmath69 .
we recall the definition of @xmath70 namely @xmath71 where @xmath72 a given constant vector and @xmath73 , the unknown vector variable .
it is easy to see that @xmath74 is a solution of ( 1.1 ) and ( 1.2 ) iff @xmath75 satisfies the burgers equation @xmath76 with initial condition @xmath77 by hopf
cole transformation , the solution can be written in the form @xmath78 as in @xcite it is easy to see that , when @xmath22 is lipschitz continuous , for each @xmath79 , except for a countable number of points of @xmath24 , the limit @xmath80 exists and is given by the formula @xmath81 where @xmath27 is a minimizer of @xmath28 and @xmath29 is given by ( 1.4 ) . here
we remark that this formula is slightly different from that of hopf @xcite .
once @xmath70 is known , ( 3.1 ) can be treated as @xmath0 scalar linear equation with discontinuous coefficient . for any function @xmath82 on
@xmath83 $ ] , @xmath84 denotes the essential supremum of @xmath82 with respect to the space variable @xmath24 keeping the time variable @xmath85 fixed .
we define the left boundary curve @xmath86 and right boundary curve @xmath87 of the support of @xmath82 where @xmath88 for all @xmath89 $ ] and @xmath90 for all @xmath91 $ ] .
it is well - known from @xcite that when the initial data is supported in @xmath92 $ ] , being solution of the inviscid burgers equation , @xmath70 have the following estimates .
there exist constants @xmath93 and @xmath94 which depend on @xmath95 and @xmath96 , such that @xmath97 we shall prove that for @xmath98 again the same estimate holds for the spread of support but the decay result is not valid in general .
first we have the following theorem .
let @xmath99 be the vanishing viscosity solution of ( 3.1 ) and ( 3.2 ) with initial data supported in @xmath92 $ ] .
let @xmath100 and @xmath101 are the support curves for @xmath102 .
then there exists a constant @xmath93 which depend on @xmath95 and @xmath103 , so that the following estimate holds for @xmath104 , @xmath105 the proof easily follows from the fact that the characteristic speed @xmath106 outside the region @xmath107 .
so in this region the characteristics connecting the points @xmath108 to a base point @xmath109 are parallel to the @xmath32-axis and hence is of the form @xmath110 and since the solution is constant along the characteristics , we have @xmath111 .
when @xmath108 is outside @xmath112 , the @xmath24 co - ordinate of the base point of the characteristic , @xmath52 , lies outside @xmath92 $ ] where @xmath113 is zero and hence @xmath114 is zero .
the proof of the theorem is complete .
@xmath62 next
we show that solution does not decay in general , by giving an example . here
we construct vanishing viscosity solution of ( 3.1 ) with initial data of the special form which is supported in a compact set , namely @xmath115 where @xmath22 is a constant and @xmath95 is a positive real number . let @xmath116 be the solution of ( 1.1 ) with the initial data ( 3.4 ) .
we shall prove the following .
let @xmath117 , then @xmath118 exists and takes the following form when @xmath119 , @xmath120 when @xmath121 , @xmath122 u_{0j } , & { \rm if } \
\frac{\sigma_0}{2}t - l < x < \sigma_0 t + l , t<\frac{-4 l}{\sigma_0}\\[.3pc ] \frac{u_{jl}}{\sigma_0}\cdot \frac{x - l}{t } , & { \rm if } \
< x < l , t<\frac{-4 l}{\sigma_0}\\[.3pc ] 0 , & { \rm if } \
x < l - ( -4 l \sigma_0 t)^{\frac{1}{2 } } , t > \frac{-4l}{\sigma_0}\\[.3pc ] \frac{u_{0j}}{\sigma_0}\cdot \frac{x - l}{t } , & { \rm if } \
l - ( -4 l \sigma_0 t)^{\frac{1}{2 } } < x < l , t > \frac{-4 l}{\sigma_0}\\[.3pc ] 0 , & { \rm if } \
x > l \end{cases}.\ ] ] when @xmath123 , @xmath124 u_{0j } , & { \rm if } \
\sigma_0 t - l < x < \frac{\sigma_0}{2}t + l , t<\frac{4 l}{\sigma_0}\\[.3pc ] \frac{u_{0j}}{\sigma_0}\cdot \frac{x+l}{t } , & { \rm if}\ -l < x < \sigma_0 t -l , t<\frac{4 l}{\sigma_0}\\[.3pc ] 0 , & { \rm if } \ x > l + ( 4 l \sigma_0 t)^{\frac{1}{2 } } , t > \frac{4l}{\sigma_0}\\[.3pc ] \frac{u_{0j}}{\sigma_0}\cdot \frac{x+l}{t } , & { \rm if } \ -l < x < l + ( 4 l \sigma_0 t)^{\frac{1}{2 } } , t > \frac{4 l}{\sigma_0}\\[.3pc ] 0 , & { \rm if}\ x < -l \end{cases}.\ ] ] to prove ( 3.5)(3.7 ) we use the formula ( 1.5 ) to get explicit solution of ( 1.1 ) and ( 3.4 ) in the form @xmath125}\d y}{\int_{-\infty}^{-l } \e^{-\frac{(x - y)^2}{2t\nu}}\d y + \int_{-l}^l \e^{-\frac{1}{\nu}\big[\frac{(x - y)^2}{2t}+\sigma_0 y\big]}\d y + \int_{l}^{\infty } \e^{-\frac{(x - y)^2}{2t\nu}}\d y}\ ] ] which can be written in the form @xmath125}\d y}{(2t\pi \nu)^{\frac{1}{2 } } + \int_{-l}^l \e^{-\frac{1}{\nu}\big[\frac{(x - y)^2}{2t}+\sigma_0 y\big]}\d y + \int_{-l}^{l } \e^{-\frac{(x - y)^2}{2t\nu}}\d y}.\ ] ] to study the limit , we rewrite this formula in a convenient form by introducing the functions @xmath126 and @xmath127 where @xmath128 we can rewrite ( 3.9 ) as @xmath129 using the asymptotic expansions of the _ erfc _ , namely , @xmath130 and @xmath131 in ( 3.10 ) and ( 3.11 ) we have the following as @xmath132 : @xmath133 @xmath134 it is straightforward to check the formulas ( 3.5)(3.7 ) using ( 3.14 ) and ( 3.15 ) in ( 3.13 ) .
when @xmath135 , ( 3.13 ) becomes @xmath136 now take the region @xmath137 . then @xmath138 and @xmath139 , and using ( 3.14 ) and ( 3.15 ) in ( 3.16 ) we get @xmath140 and hence we have @xmath141 in the region @xmath142 , @xmath143 , @xmath144 thus we get @xmath145 in the region @xmath146 , we have @xmath147 and @xmath148 and hence it follows that @xmath149 this completes the proof of the theorem for @xmath135 .
the case @xmath150 is similar and is omitted .
thus if we take the initial data ( 3.4 ) which has compact support the decay of the vanishing viscosity solution @xmath151 depends on the initial speed @xmath152 .
if @xmath135 , the vanishing viscosity solution does not decay .
indeed @xmath153 on the other hand , from the above theorem it follows that for the case @xmath154 , the solution decays , namely @xmath155 we conclude with a remark on the solution for ( 3.1 ) with the riemann type initial data @xmath156 where @xmath157 and @xmath158 are constants for @xmath159 .
let @xmath160 and @xmath161 .
then the vanishing viscosity solution for the riemann problem ( 3.1 ) and ( 3.17 ) takes the following form @xcite . when @xmath162 , @xmath163 \frac{u_{jr}-u_{jl}}{\sigma_r-\sigma_l}\cdot \frac{x}{t } + \frac{u_{jl } \sigma_r - u_{jr } \sigma_l}{\sigma_r - \sigma_l } , & { \rm if } \
\sigma_{l } t < x < \sigma_{r}t.\\[.2pc ] u_{jr } , & { \rm if } \
x\geq\sigma_rt\end{cases}\ ] ] when @xmath164 , @xmath165 u_{jr } , & { \rm if } \
x > \sigma\cdot t.\end{cases}\ ] ] when @xmath166 , @xmath167 \frac{u_{jl}+u_{jr}}{2 } & { \rm if } \ x = \frac{\sigma_l + \sigma_r}{2}\cdot t\\[.3pc ] u_{jr } , & { \rm if } \ x>\frac{\sigma_l + \sigma_r}{2 } \cdot t \end{cases}.\ ] ]
this work is supported by a grant ( no .
26012 ) from the indo - french centre for the promotion of advanced research , ifcpar ( centre franco - indien pour la promotion de la recherche avancee , cefipra ) , new delhi . lefloch
p g , an existence and uniqueness result for two nonsrtictly hyperbolic systems , in : nonlinear evolution equations that change type ( eds ) barbara lee keyfitz and michael sheaer , i m a volumes in mathematics and its applications ( 1990 ) ( springer - verlag ) vol . 27 , pp . 126139 | in this paper we study the asymptotic behaviour of solutions of a system of @xmath0 partial differential equations .
when @xmath1 the equation reduces to the burgers equation and was studied by hopf .
we consider both the inviscid and viscous case and show a new feature in the asymptotic behaviour .
[ theore]*theorem * [ theore]remark [ theore]proposition [ theore]lemma [ theore]definition [ theore]corollary [ theore]example |
polyominoes are two dimensional objects which are originally rooted in recreational mathematics and combinatorics .
they have been widely discussed in connection with tiling problems of the plane . typically , a polyomino is plane figure obtained by joining squares of equal sizes , which are known as cells . in connection with commutative algebra ,
polyominoes are first discussed in @xcite by assigning each polyomino the ideal of inner 2-minors or the _ polyomino ideal_. the study of ideal of @xmath0-minors of an @xmath1 matrix is a classical subject in commutative algebra .
the class of polyomino ideal widely generalizes the class of ideals of 2-minors of @xmath1 matrix as well ass the ideal of inner 2-minors attached to a two sided ladder .
let @xmath2 be a polyomino and @xmath3 be a field .
we denote by @xmath4 , the polyomino ideal attatched to @xmath2 , in a suitable polynomial ring over @xmath3 . the residue class ring defined by @xmath4 denoted by @xmath5 $ ] .
it is natural to investigate the algebraic properties of @xmath4 depending on shape of @xmath2 .
the classes of polyominoes whose polyomino ideal is prime has been discussed in many papers . in @xcite
it is shown that polyomino ideal of convex polyominoes are prime .
in @xcite they introduced the class of _ balanced _ polyominoes and proved that polyomino ideal of balanced polyominoes are prime . later in @xcite
it is shown that simple polyominoes are balanced . in @xcite , independently from @xcite , it is proved that polyomino ideal of simple polyominoes are prime by identifying them with toric rings of edge rings of graphs .
very recently , in @xcite it is shown that polyomino ideal of polyomino of the shape `` rectangle minus convex '' is prime by using localization argument on the attached rings of those polyominoses . in this paper
, we identify the toric rings of polyomino ideals for the special class of nonsimple polyominoes , `` rectangle minus rectangle '' .
first we recall some definitions and notation from @xcite . given @xmath6 and @xmath7 in @xmath8 we write @xmath9 if @xmath10 and @xmath11 .
the set @xmath12=\{c\in\nn^2\:\ ; a\leq c\leq b\}$ ] is called an _
interval_. if @xmath13 and @xmath14 , then the elements @xmath15 and @xmath16 are called _ diagonal _ corners and @xmath17 and @xmath18 are called _ anti - diagonal _ corners of @xmath12 $ ] .
an interval of the from @xmath19 $ ] is called a _ cell _ ( with left lower corner @xmath15 ) . the elements ( corners ) @xmath20 of @xmath21 $ ]
are called the _ vertices _ of @xmath22 . the sets @xmath23 and @xmath24
are called the _ edges _ of @xmath22 .
we denote the set of edge of @xmath22 by @xmath25 .
let @xmath26 be a finite collection of cells of @xmath8 .
the the vertex set of @xmath26 , denoted by @xmath27 is given by @xmath28 .
for each vertex @xmath29 in @xmath30 , we write @xmath31 where @xmath32 is the first and @xmath33 is the second coodinate of @xmath29 . the edge set of @xmath26 , denoted by @xmath34
is given by @xmath35 .
let @xmath22 and @xmath36 be two cells of @xmath26 . then @xmath22 and @xmath36 are said to be _ connected _ , if there is a sequence of cells @xmath37 of @xmath26 such that @xmath38 is an edge of @xmath39 for @xmath40 .
if in addition , @xmath41 for all @xmath42 , then @xmath43 is called a _ path _ ( connecting @xmath22 and @xmath36 ) .
the collection of cells @xmath26 is called a _ polyomino _ if any two cells of @xmath26 are connected , see figure [ polyomino ] .
let @xmath26 be a polyomino , and let @xmath3 be a field .
we denote by @xmath44 the polynomial ring over @xmath3 with variables @xmath45 with @xmath46 .
we write for @xmath47 , we may write @xmath48 if needed . following @xcite a @xmath49-minor @xmath50 is called an _ inner minor _ of @xmath26 if all the cells @xmath51 $ ] with @xmath52 and @xmath53 belong to @xmath26 . in that case
the interval @xmath54 $ ] is called an _ inner interval _ of @xmath26 . the ideal @xmath55 generated by all inner minors of @xmath26
is called the _ polyomino ideal _ of @xmath26 .
we also set @xmath56=s / i_\pc$ ] .
for each interval @xmath12 $ ] we regard the polyomino @xmath57 in the obvious way .
we give theorem [ boxbox ] that identifies a toric ring for some specified cases of nonsimple polyominoes which is obtained by subtracting a rectangle from a bigger rectangle .
let @xmath69 $ ] be the subring of a polynomial ring + @xmath70 $ ] where @xmath71 we define the surjective ring homomorphism @xmath72 \rightarrow t$ ] by setting @xmath73 .
for the sake of convenience , we write @xmath74 in case it does not make readers confused .
we call the _ toric ideal _
@xmath75 first it is easy to see that @xmath77 ker@xmath78 . we show that every quadratic binomials belong to @xmath79 is a generator of @xmath4 , namely an inner 2-minor in @xmath2 and every binomials of higher degree belongs to @xmath4 .
first , suppose that @xmath80 is a quadratic binomial belongs to @xmath79 with @xmath81 .
then either @xmath82 or @xmath83 holds .
thus , the quadratic binomial @xmath84 is associated with an interval .
now we show that the interval is an inner interval in @xmath2 .
assume that the interval spanned by @xmath85 has @xmath32 and @xmath33 as diagonal corners .
also , we may assume that@xmath86 and @xmath87 hold .
suppose @xmath88 $ ] is not an inner interval in @xmath2 .
then @xmath89 } \subset \p_{[v_1,v_2]}$ ] can not be happened because otherwise , from our definition of @xmath90 , one has @xmath91 .
therefore we may assume that one of @xmath92 $ ] , @xmath93 $ ] , @xmath94 $ ] or @xmath95 $ ] is not an inner interval in @xmath2 .
then from our definition of labelling we can not have @xmath96 .
hence every degree 2 binomial in @xmath79 is a generator of @xmath4 .
now we show that each binomial of higher degree belongs to @xmath4 .
let @xmath97 ker@xmath78 be a homogeneous binomial .
let @xmath98 be the set of vertices such that @xmath99 appear in @xmath100 and @xmath101 the vertices appear in @xmath102 .
now we show that if we have @xmath108 such that @xmath109 , then we are done in this case .
assume that we have a vertex @xmath110 in @xmath111 such that @xmath112 and @xmath113 .
since @xmath107 belongs to the kernel , we have 2 vertices , say @xmath32 and @xmath33 in @xmath114 such that @xmath115 and @xmath116 .
let @xmath117 be the rest corner of the interval spanned by @xmath32 , @xmath33 and @xmath110 .
our rule of labelling show that the interval is an inner interval in @xmath2 see figure [ fig : side ] to see the situation .
now we assume every vertex in @xmath121 and @xmath122 are labelled 1 or 2 .
for @xmath123 , we can find a vertex @xmath32 and @xmath33 in @xmath124 where @xmath125 and @xmath126 .
if @xmath127 or @xmath128 , then using the same formula as above , we can apply induction .
now we assume that @xmath129 and @xmath130 . by using similar argument
, one has @xmath131 such that @xmath132 .
since @xmath133 holds , we have a vertex @xmath131 such that @xmath134 .
see figure [ situation ] for this situation . then the interval spanned by @xmath110 and @xmath135 is an inner minor of @xmath2 .
let @xmath136 be the other corners of this interval . then
, @xmath137 notice that one of @xmath138 or @xmath139 holds .
assume @xmath140 .
then the interval spanned by @xmath141 and @xmath142 has @xmath32 as a corner vertex .
suppose that @xmath143 is the rest corner vertex of this interval .
then , we have @xmath144 since @xmath145 and @xmath146 are inner minors of @xmath2 and since degree of @xmath147 is less than degree of @xmath107 , we can apply induction on degree of @xmath107 to complete the proof .
in this paper we identify toric rings of polyominoes which are of the form `` rectangle minus rectangle '' .
in general it is interesting but not so easy to identify toric ring for a given binomial prime ideal .
the wider class of polyominoes which can be discussed about its toric ring is `` rectangle minus convex '' , because it is proved in @xcite that their polyomino ideals are prime . | it is known that toric ring of a simple polyomino is ring homomorphic to a edge ring of a weakly chordal bipartite graph . in this paper
we identify the attached toric rings of nonsimple polyominoes which are of the form `` rectangle minus rectangle '' .
[ multiblock footnote omitted ] |
in this appendix , we provide the details of a description of gaussian states and their relation to states capturing massive particles . as described in the main text , the gaussian state corresponding to measured second moments
is given by @xmath66 this maximum entropy state is achieved by the generalised gibbs ensemble @xmath67 where @xmath68 is the usual partition sum and the @xmath69 are determined by demanding that @xmath70 this state is a gaussian quantum state . the optimality can easily be seen by considering the relative entropy distance @xmath71 where the term on the right hand side simplifies to @xmath72 for a gaussian state , and observing that this is a non - negative functional .
this expression can be calculated using the bosonic partition sum @xmath73 with this , we straightforwardly haver @xmath74 and hence @xmath75
to conclude our estimate , we need to solve a convex minimisation problem subject to the measurement results @xmath76 where , for convenience , we have suppressed the index @xmath36 labelling the mode in question .
the lagrangian to this problem is given by @xmath77 where @xmath78 are lagrange multipliers .
the extremal points of this lagrangian are of the form @xmath79 for fixed @xmath80 .
thus the lagrange dual function is given by @xmath81 invoking lagrange duality , the bound of the main text can hence br proven . | ultra - cold atoms in optical lattices provide one of the most promising platforms for analog quantum simulations of complex quantum many - body systems .
large - size systems can now routinely be reached and are already used to probe a large variety of different physical situations , ranging from quantum phase transitions to artificial gauge theories . at the same time , measurement techniques are still limited and full tomography for these systems seems out of reach . motivated by this observation , we present a method to directly detect and quantify to what extent a quantum state deviates from a local gaussian description , based on available noise correlation measurements from in - situ and time - of - flight measurements .
this is an indicator of the significance of strong correlations in ground and thermal states , as gaussian states are precisely the ground and thermal states of non - interacting models .
we connect our findings , augmented by numerical tensor network simulations , to notions of equilibration , disordered systems and the suppression of transport in anderson insulators .
ultra - cold atoms in optical lattices provide one of the most prominent architectures to probe the physics of interacting many - body systems .
the parameters of the hamiltonians emerging in this fashion allow to explore a wide range of physical phenomena in and out of equilibrium @xcite .
they are one of the most promising platforms for realising quantum simulations and showing signatures of outperforming classical computers for certain problems @xcite . using state - of - the - art techniques , large scale systems with several thousand atoms
can be controlled @xcite ; in fact , even states with specific initial configurations and atoms aligned on largely arbitrary shapes can be realised @xcite . moreover , by modulating the optical lattice in time or by altering its geometry , a wide range of complex physical settings can be explored , ranging from probing quantum phase transitions @xcite to realising instances of artificial gauge theories @xcite .
while a large degree of control over these platforms has been achieved , at present , it is still true that the measurement capabilities are limited in practice .
this appears particularly relevant in the context of quantum simulations , where the result of the simulation has to be read out from the physical experiment .
it seems clear that full quantum state tomography is infeasible , both for limits in the availability of measurement prescriptions as well as due to the unfavourable scaling of the tomographic effort with the system size .
suitable combinations @xcite of tensor network tomography @xcite and compressed sensing schemes @xcite suggest a way forward towards achieving tomographic knowledge , but at present such ideas have not been realised yet . in the light of these obstacles
, it seems imperative to focus the attention to developing tools directly that detect relevant properties of the quantum state , rather than trying to capture the full density operator which gives rise to information that is often not needed . among those , entanglement features come to mind that contain valuable information about a quantum state @xcite , or notions of non - classicality that can be directly detected @xcite .
similarly , it is of interest to identify and quantify to what extent the state realised corresponds to a ground or thermal state of an interacting model , and hence to what extent the state deviates from a gaussian state .
as one of the main promises of the field of ultra - cold atoms is to precisely study interacting quantum many - body models deviating from non - interacting theories , this type of information is highly relevant , precisely probing to what extend an observation is compatible with a non - interacting model not exhibiting strong correlations . in this work ,
we introduce a new scheme that can be used to directly estimate the local gaussianity of a state , based solely on second and fourth moments of particle number measurements .
we apply this tool to the specific context of ultra - cold atoms and show that noise correlations in in - situ measurements are already sufficient for its calculation . in this way
, we build upon and relate to the ideas of refs .
@xcite , but deliver an answer to the converse task : we do not show how interacting models are reflected in noise - correlations , but ask how data can be used to unambiguously witness such deviations from non - interacting models . interacting many - body quantum systems
are exceedingly hard to capture and describe in terms of classical parameters .
non - interacting models
models that have gaussian ground and thermal states are an exception to this rule , in that their description complexity is low .
they are a paradigmatic class of states both for fundamental questions in quantum information as well as for finding ground states of condensed matter models , such as the interaction - free bose - hubbard model @xcite . in this sense
, they are states that do not exhibit the intricate structure of interacting quantum many - body models . again , since one of the main promises of the field is to address such interaction effects , it seems important to have tools at hand to directly detect a deviation from gaussianity .
the bosonic gaussian states discussed here are characterised by the second moments collected in the correlation matrix @xmath0 with entries @xmath1 where @xmath2 denote the canonical bosonic creation and annihilation operators , @xmath3 , for an @xmath4 mode system . throughout this work
, we investigate massive bosons leading to a situation in which all @xmath5 for all @xmath6 .
any such correlation matrix satisfying @xmath7 can be diagonalised with a unitary @xmath8 as @xmath9 , reflecting a mode transformation @xmath10 preserving the bosonic commutation relations , where @xmath11 are the transformed modes .
one immediately finds @xmath12 where @xmath13 denotes the von - neumann entropy and @xmath14 denotes the set of density matrices .
in addition , by invoking the pinching inequality @xcite , one finds that it is already sufficient to fix the diagonal entries of the correlation matrix @xmath15 for @xmath16 @xmath17 following from this , given the diagonal elements of the correlation matrix in the momentum representation of the modes , the following gaussian state is uncorrelated over the individual modes and given by ( [ gge ] ) @xmath18 where @xmath19 is corresponding to @xmath15 ( see appendix [ app_gaussian ] ) .
these states can be viewed as an instance of a generalised gibbs ensemble @xcite .
importantly for the context at hand , such gaussian states also play a prominent role in the setting of optical lattices , where they can , for example , be used to capture the superfluid ground state . in recent years , research on cold atoms in optical lattices has progressed significantly , by now allowing for unprecedented control of interacting quantum many - body systems , involving several thousands of atoms .
relying on recent experimental advances , the position of individual atoms can be tracked using single site addressing @xcite .
using such techniques , local expectation values of the particle number as well as density - density correlations can be resolved .
these local measurements have already provided important insight into the out - of - equilibrium dynamics of quantum many - body systems @xcite and allowed to access their microscopic properties .
the main result of this work , which we detail subsequently , is to identify tools to detect a deviation from gaussianity reflecting a non - interacting system based on particle number measurements , building upon refs .
@xcite . for this
, we begin with a clarifying discussion to what extent states encountered in optical lattices can be gaussian .
states in optical lattices describe massive particles . for that reason ,
the particle number in each experimental run is fixed .
this , however , implies that the full state is not gaussian as we show below .
the best example for this is the perfect superfluid state , corresponding to the ground state of the 1d free hopping hamiltonian @xmath20 using a chemical potential @xmath21 to maintain the expected particle number constant , thermal states of this hamiltonian take the form @xmath22 where @xmath23 denotes the inverse temperature .
this ensemble is gaussian , as it is the exponential of a quadratic expression of creation and annihilation operators .
it , however , only fixes the particle number on average . in order to fix the actual particle number , meaning to ensure that the state lives on a fixed particle number sector also all higher moments have to be included in the ensemble
however , already including a fixed variance with a lagrange parameter @xmath24 @xmath25 results in a state that is no longer strictly gaussian .
nevertheless , the superfluid state can be thought of being gaussian in one important sense .
local particle number measurements are indistinguishable from a gaussian state for sufficiently large system sizes is is not that obvious .
thus , many measurements of the state can be captured using a simple gaussian description . in the following , we present a general method that allows us to estimate the local deviation from gaussianity of a state , based solely on measurements of second and fourth moments , which will afterwards be applied to states capturing optical lattices . in order to determine the local deviation from gaussianity of states in optical lattices , we first describe how the distance of the global state to the manifold of gaussian states can be captured and estimated relying only on second and fourth moments .
we then use this insight to define a local deviation from gaussianity and evaluate it for paradigmatic bosonic models . to quantify the deviation of a state from gaussianity
, we use the relative entropy as a natural quantity with a precise statistical interpretation also known as kullback - leibler divergence .
the relative entropy between two states @xmath26 and @xmath27 is defined as @xmath28 this quantity provides the asymptotic statistical distinguishability of @xmath26 from @xmath27 in the situation of having available arbitrarily many copies of the state @xcite .
based on the relative entropy , we define the global deviation from gaussianity of a state as @xmath29 where @xmath30 denotes the set of all gaussian states .
this can be seen as a measure of the strong correlations present in the state , as quantifying the statistical deviation from a state that could have been the ground or thermal state of a non - interacting model .
as it turns out , the minimum is always achieved for the gaussian state @xmath27 with the same correlation matrix as @xmath26 @xcite .
we denote this special gaussian state by @xmath31 .
the global deviation from gaussianity can be lower bounded by relying only on local measurement data , which will be performed in the following .
we begin by using the additivity of the relative entropy which allows us to describe the problem in the symplectic eigenbasis of @xmath31 and decompose the estimate into a problem involving individual eigen - modes @xcite @xmath32 where @xmath33 denotes the unitary transformation in hilbert space ( metaplectic representation ) reflecting the moment transformation into the eigen - modes of @xmath0 , @xmath34 and @xmath35 denotes the reduction to the @xmath36-th mode . making use of the additivity of the relative entropy for each individual mode , we can use the deviation from gaussianity of @xmath31 and rewrite the estimate in terms of entropies @xcite @xmath37 these single mode entropy estimates are a major simplification compared to the original problem .
naturally calculating or measuring the entropy of @xmath38 is still not a task that can be performed efficiently or effectively .
rather , we solely rely on measured fourth moments @xmath39 and calculate the smallest possible distance to the manifold of gaussian states compatible with this data . in this way , we are able to compute a minimal deviation from gaussianity , thus showing that the state is , in this precise sense , strongly correlated .
we therefore relax the problem as follows @xmath40 where @xmath41 is the maximum entropy state compatible with the second and fourth moments . using schur s theorem @xcite , which states that the ordered eigenvalues of a matrix majorise the ordered diagonal entries @xmath42 and the fact that the von - neumann entropy is schur convex , the entropy of this state can be upper bounded by only considering the diagonal @xmath43 the second index labelling the main diagonal elements .
this allows for an efficient solution of the optimisation problem using lagrange multipliers ( see appendix [ app_singlemode ] ) .
thus , we have seen that the global deviation from gaussianity defined in eq . can be lower bounded by using second and fourth moments in the symplectic eigenbasis of the state .
based on this insight , we turn back to states describing massive particles in optical lattices . there
, the full state can not be reconstructed , which necessarily implies that only important features of the state may be addressed .
here we focus on a particularly simple quantity that only relies on measuring the particle number on a single site , which is accessible experimentally using single - site addressing @xcite . using such data
, we define the _ local deviation from gaussianity _ on site @xmath44 of a state as @xmath45 where @xmath46 is the particle number operator on site @xmath44 and @xmath47 denote its experimentally measured expectation value and variance .
this local deviation from gaussianity is not only comparably easily measurable , but more importantly also yields relevant information about the quantum system at hand . as discussed above
, the superfluid state is locally gaussian in the sense that the quantity in eq .
vanishes for large enough system sizes . in this way , calculating the local deviation from gaussianity , which captures to what extend onsite particle measurements are compatible with a gaussian state , is a natural way to quantify the distance to a perfect superfluid state .
moreover , for the important case of non - interacting particles , it can also be used to identify the suppression of particle propagation due to disorder .
both these applications are elaborated upon below . in the following , we present numerical results for bosonic models commonly encountered in optical lattices .
we begin with the attractive bose - hubbard model @xmath48 where we have chosen the hopping strength equal to one and denote the interaction strength with @xmath49 .
the ground state for @xmath50 is the idealised superfluid state introduced above and thus locally gaussian in our sense , corresponding to a local deviation from gaussianity of zero .
in contrast , when the interaction strength is increased , the system becomes strongly correlated and the size of the local deviation from gaussianity should increase .
we have confirmed that this behaviour is indeed encountered numerically and find an almost linear relation between the local deviation from gaussianity and the interaction strength ( see fig .
[ fig : ground ] ) . in this way
, one could even see the local deviation from gaussianity as an experimental probe to directly measure the interaction strength based solely on ground state particle number fluctuations .
and interaction strength @xmath49 .
the results are obtained with exact diagonalisation using periodic boundary conditions on @xmath51 lattice sites . ]
another setting in which the local deviation from gaussianity is intriguing to investigate is the evolution of free systems , where it can be used as an indicator of disorder and the concomitant suppression of transport .
it is known that non - interacting systems which exhibit transport in a suitable sense evolve in time in a way that the states tend to locally gaussian states following out of equilibrium dynamics @xcite , a feature that is true in surprising generality @xcite .
thus , in this setting , the precise initial conditions are forgotten over time and local expectation values can be captured using only the second moments of the initial state , then fully determining local expectation values .
this applies in particular to the free hopping hamiltonian introduced in eq . .
a paradigmatic setting for this is given by an initial product state with one particle on every second site , which can be experimentally prepared employing optical super - lattices @xcite .
this initial state is the ground state of an infinitely strongly interacting bose - hubbard model and it is thus far from being locally gaussian , as described above . during time evolution that has transport , however ,
the particles distribute evenly over the lattice , thus moving towards the manifold of locally gaussian states . for disordered systems
, transport is strongly suppressed and the distribution of particles over the lattice thus does not take place . for concreteness ,
let us consider a simple 1d hopping model @xmath52 with local potentials @xmath53 drawn uniformly from some interval @xmath54 $ ] .
for @xmath55 the model reduces to the free hopping and local gaussification thus takes place . in contrast , when randomness is present , transport breaks down , resulting in a positive local deviation from gaussianity even for long times ( see fig .
[ fig : time ] ) . sites .
the initial state is a charge - density wave corresponding to a fock state with one atom on every second lattice site and the evolution is governed by the hamiltonian in eq . with @xmath53 drawn randomly from the interval @xmath54 $ ] for three different values of the disorder strength .
shown is the average over @xmath56 disorder realisations .
initially , the local state is very non - gaussian . for the translationally invariant system showing transport ,
information spreads through the lattice and the local measurements become compatible with a fully gaussian description of the state , that is , the system dynamically gaussifies .
in contrast , in the disordered case , transport is strongly suppressed and the non - gaussianity remains locally visible . ] the same approach discussed here for local deviation from gaussianity can be applied to global properties of the lattice .
in fact , it can be applied to any modes that are defined by a mode transformation @xmath57 in particular , it is applicable to time - of - flight measurements . if the quantum state is not translationally invariant due to the presence of a harmonic trap , an average deviation from gaussianity over the system size is then directly detected in this way . in time - of - flight measurements with finite accuracy of the camera pixels ,
effectively the diagonal elements @xmath58 of @xmath59 are measured , more commonly expressed as @xmath60 as a function of @xmath61 , where @xmath62 is the time of flight and @xmath63 is a constant derived from the mass and the lattice constant of the optical lattice .
the fourth moments of the same modes defined by @xmath64 are accessible as @xmath65 and contained in the very same images from the laboratory , merely by computing higher moments , following a prescription of ref .
@xcite .
let it be stressed again that in contrast to this reference , we aim for and provide a direct detection of correlations based solely on second and fourth moments . in this work ,
we have introduced and elaborated on a method of directly detecting local deviation from gaussianity of quantum many - body systems . in this way
, we have identified a way of witnessing the deviation from the ground or thermal state of a non - interacting model , hence directly observing strong correlations present in the state .
we did so by making use of ideas of convex optimisation .
more conceptually speaking , the mindset that we advocated here is to make use of currently available tools to directly detect properties relevant for a research question at hand . given the simplicity of the bounds , it constitutes a relevant further step to equip the bounds and estimates presented here with precise statistical confidence regions .
this complements a more conventional approach in which a situation sharing some feature is modelled or classically simulated , and the predictions compared with those of data from measurements .
it is the hope that the present work can contribute to the further development of such tools of certification for the study of many - body models , particularly relevant for the partial certification of the functioning and read out of analog quantum simulators .
10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , , &
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ .
_ _ * * , ( ) .
_ et al . _ . . ,
& . _
_ * * , ( ) .
_ et al . _ .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
, , , & .
_ _ * * , ( ) . , & .
_ _ * * ( ) . , & .
_ _ * * , ( ) . & _ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
, , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . .
.
_ _ ( , ) . , , & .
_ _ * * , ( ) . , & .
_ _ * * , ( ) . , , & . _ _ ( ) .
_ et al . _ . _
* * , ( ) .
_ et al . _ . _
_ * * , ( ) . , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . .
_ _ * * , ( ) .
_ _ ( , , ) . .
& . _
_ * * , ( ) . , & .
_ _ * * , . & .
_ _ * * , ( ) . , , , & . .
_ et al . _ .
_ _ * * , ( ) .
we sincerely thank m. ohliger and u. schneider for stimulating discussions .
we would like to thank the eu ( aqus , raquel ) , the erc ( taq ) , the bmbf ( q.com ) , and the dfg ( ei 519/9 - 1 , ei 519/7 - 1 ) for support .
car also acknowledges the support of the freie universitt berlin within the excellence initiative of the german research foundation . |
in addition to the known particles and interactions of the standard model ( sm ) , there are many compelling reasons to consider a ` hidden world ' of particles .
we define a hidden sector generically as any set of particles , in addition to those of the sm , that are not charged under any of the sm gauge groups .
the possibilities for such models are endless , and if they have no testable consequences then such speculation has little scientific merit .
however , there do exist a limited number of interactions that could exist between a hidden sector and the sm that are renormalisable and relevant or marginal , i.e. corresponding to terms in the lagrangian with dimension @xmath2 4 and which therefore are not suppressed below some energy scale .
thus , it is reasonable to identify and explore such possibilities as potential windows into a hidden sector .
one possibility is kinetic mixing of abelian gauge fields of the hidden sector and the sm @xcite , where the resulting phenomenology overlaps with standard @xmath3 physics @xcite .
a second possibility is a renormalisable interaction between scalar fields in the two sectors .
this mixing would be between the sm higgs boson , which may have already been discovered @xcite , and an exotic singlet scalar state that obtains a vacuum expectation value .
this possibility therefore corresponds to an extended higgs sector , with a new scalar that may be responsible for spontaneous symmetry breaking in the hidden sector . for concreteness , we study a minimal ` toy model ' of this type the hidden abelian higgs model @xcite which involves the addition of a new scalar singlet charged under a @xmath4 gauge symmetry .
the phenomenological consequences of this model for lhc physics have been well studied , but there are significant regions of the parameter space in which the new higgs particle would be very difficult to detect at the lhc if its mass is @xmath0 1 tev @xcite . the purpose of the present work , therefore , is to investigate the possibility of detecting such a particle at the compact linear collider ( clic ) , a future electron - positron collider experiment currently under discussion @xcite .
this experiment would have some advantages for detecting such a particle when compared to the lhc , so this possibility is well worth exploring .
we present our key findings on the basis of a parton level analysis for the recoil mass spectrum from @xmath5 production in sections [ subsec : elecrecoil ] , the @xmath6 signal in section [ subsec : zz ] , and finally the @xmath7 signature in [ subsec : hh ] .
we begin with a review of the hidden abelian higgs model . in this model
the hidden sector contains a complex scalar singlet @xmath8 which is charged under a @xmath9 gauge symmetry .
the higgs lagrangian is @xmath10 with @xmath11 , @xmath12 , @xmath13 and @xmath14 all positive ( while @xmath15 can take either sign ) . in this case , the @xmath9 gauge symmetry is spontaneously broken by @xmath8 taking a particular vacuum expectation value ( vev ) ( @xmath16 ) .
the resultant massless goldstone mode gets absorbed by the vector boson corresponding to the abelian gauge symmetry , as in the sm . in the unitary gauge which eliminates the unphysical goldstone modes ,
the scalar fields can be written @xmath17 where v(@xmath18 gev ) and @xmath19 are the vevs of @xmath20 and @xmath8 , while @xmath21 and @xmath22 are the massive modes of these fields .
however , the mixing term @xmath23 means that the physical mass eigenstates @xmath24 , @xmath25 of the theory are a linear combination of the two @xmath26 where we choose @xmath27 , with @xmath28 the higgs masses and mixing are subject to a number of constraints .
precision electroweak constraints and direct searches play central roles in constraining what the parameter space can be .
the model has four free parameters .
starting with two key observables @xmath29 and @xmath30
one is left with two parameters that can be chosen freely .
an analysis of the parameter space of the theory @xcite has shown that large regions are compatible with the theoretical constraints , including the regions which are explored in this paper . for @xmath31 ,
which we assume throughout this study , the decay @xmath32 is possible . the partial width for this decay is given by the tree level formula @xmath33 where @xmath34 is the coupling associated with the @xmath35 term in the lagrangian , the expression for which can be found in @xcite .
@xmath25 can also decay into the sm particles , predominantly @xmath36 , @xmath37 or @xmath38 , with a width @xmath39 , where @xmath40 is the standard tree level result for a sm higgs of mass @xmath29 .
@xmath25 could , in principle , also decay to hidden sector particles with a hidden width @xmath41 .
this possibility has been discussed in @xcite , and in section [ subsec : elecrecoil ] we briefly touch on a clic search channel which could be useful if the hidden width is significant .
however , in this paper we assume that hidden decays of @xmath25 are kinematically forbidden ( i.e. the mass of the hidden sector products exceeds @xmath29 ) and so only the sm and @xmath32 decay channels contribute to the total width .
the @xmath42 suppression of widths is central to this study .
a sm higgs resonance becomes so wide once its mass is in the trans - tev region that it becomes impossible to find and it can no longer be understood as a particle .
this suppression allows us to keep the @xmath25 resonance reasonably narrow despite its large mass , and therefore to exploit standard search channels which have already been considered for the heavy ( but sub - tev ) sm higgs case .
clic is a proposed linear @xmath43 collider presently under discussion @xcite .
it is proposed to have a nominal center of mass energy up to 3 tev , and a design luminosity of @xmath44 @xmath45s@xmath46 which should result in more than @xmath47 fb@xmath46 of integrated luminosity per year at peak design .
lepton collision events have the advantage of being ` cleaner ' than at a hadron collider , however , the added complication of strong beamstrahlung at a tev - scale lepton collider presents its own set of issues .
beamstahlung is the emission of hard photons from the colliding bunches , a consequence of the bending of particle trajectories in the strong electromagnetic fields which results from the small bunch sizes required in a high luminosity linear collider @xcite .
the effects which are most relevant to our study are the resulting @xmath48 hadron background , and the spread in the center of mass energies of @xmath43 collisions
. collisions between beamstrahlung photons result in hadron production focussed mainly in the forward region of the detector , ( @xmath49 100 mrad ) , but also with energy deposition per event @xmath50 50 gev in the region @xmath51 280 mrad relevant to the hadronic calorimeters @xcite .
this has the potential to impact jet reconstruction and missing energy measurements , and proper evaluation of this background requires a detector simulation .
we have made no attempt to account for this backround in our study , however , we expect its impact to be modest ( e.g. , see p.553 of @xcite ) .
more significant is the effect of beamstrahlung on the luminosity spectrum ; electrons loose energy due to photon emission , and the effects on a colliding pair are correlated which makes a parametrisation of the spectrum difficult .
we account for this by making use of a luminosity file generated from the program guineapig @xcite , which simulates the beam - beam interaction and produces a list of pairs of particle energies which can be used in a monte carlo simulation on an event - by - event basis .
heavy higgs production at clic proceeds predominantly via vector boson fusion , either by @xmath37 fusion with final state @xmath43 , or @xmath36 fusion with final state @xmath52 .
the @xmath53 ` higgstrahlung ' process , which dominates at a low energy lepton collider like the international linear collider ( ilc ) , gets weaker with increasing center of mass energy such that at clic for a 1 tev sm higgs the higgstrahlung production cross section is @xmath54 1 fb compared with 8.4 fb for @xmath37 fusion and 79 fb for @xmath36 fusion @xcite .
although @xmath55 higgs boson is somewhat meaningless concept in the sm , due to the strong , non - perturbative growth of its couplings at that high mass , the computation of the cross - section at leading order is still useful for our purposes .
this is because the state we are interested in the lightly mixed - in singlet higgs boson has couplings that are small in comparison and perturbative due to the small mixing angle . in the following sections we investigate three search strategies , focussing on that described in section [ subsec : zz ] .
all signal processes are simulated at tree level using the monte carlo event generator pythia 6.4.24 @xcite , while background processes are simulated using comphep 4.4 @xcite .
the luminosity spread due to beamstrahlung is included in the pythia simulations using calypso @xcite .
the beamstrahlung tool available in comphep is suitable for a low energy lepton collider , but the yokoya - chen approxiation which is used breaks down at high energies and so is unsuitable for our study @xcite .
the backgrounds plotted in section [ subsec : zz ] do not include the effects of the luminosity spread . as we will discuss later
, we consider this as a small systematic error in our subsequent analysis .
all signals are simulated in pythia using calypso and so include the effects of beamstrahlung ( except where explicitly stated otherwise ) .
one inclusive search strategy is to look for recoiling high energy electrons from the @xmath37 fusion process , which are peaked in the forward ( small polar angle ) region .
this has the advantage of summing over all possible higgs decays , even invisible ones , and so provides a rather model - independent search channel @xcite .
it suffers from relying on the smaller @xmath37 fusion cross section , and from the fact that instrumentation for electron tracking and tagging can only cover angles @xmath0 100 mrad and therefore can pick up only the tail of the electron distribution which peaks at @xmath56 50 mrad ( see left panel of figure [ fig : recoil ] ) .
events that result in both the outgoing electron and the positron taking an angle @xmath51 120 mrad represent only 12% of the total production cross section .
the electron recoil mass is the invariant mass of the decay products against which the @xmath43 pair recoil assuming the collision occurs at the nominal center of mass energy , defined as @xmath57 a higgs resonance would potentially show up as a resonant peak in this electron recoil mass spectrum , centered at @xmath29 .
we plot an electron recoil mass spectrum for @xmath58 = 1 tev , @xmath59 = 0.3 in figure [ fig : recoil ] , both excluding ( solid black line ) , and including ( dashed blue line ) the effects of beamstrahlung on the luminosity spectrum .
cuts applied are @xmath60 120 mrad and @xmath61 700 gev .
total cross sections for a range of points in the parameter space are presented in table [ tab:1 ] .
the width is calculated under the assumption that @xmath25 decays only into sm particles , though as explained earlier this channel could be used to search for an invisibly decaying higgs .
it can be seen from table [ tab:1 ] that this channel is promising for a substantial region of the parameter space ( @xmath62 at @xmath63 1 tev , @xmath64 at @xmath65 2 tev ) , so long as backgrounds are small , as is expected .
it should be noted that for a 600 gev sm higgs boson , a consistency - check calculation using this method gives a cross section which is a factor of 4 smaller than given in previous work @xcite , apparently using the same cuts , and the reason for this discrepancy is unknown .
= 1 tev and @xmath59 = 0.3 .
the dashed blue ( solid black ) line shows the mass spectrum accounting ( not accounting ) for the effects of beamstrahlung on the luminosity spectrum .
note that the differential cross section must be multiplied by the bin width ( 20 gev in the right plot ) , to get the cross section per bin.,title="fig:",scaledwidth=50.0% ] = 1 tev and @xmath59 = 0.3 .
the dashed blue ( solid black ) line shows the mass spectrum accounting ( not accounting ) for the effects of beamstrahlung on the luminosity spectrum .
note that the differential cross section must be multiplied by the bin width ( 20 gev in the right plot ) , to get the cross section per bin.,title="fig:",scaledwidth=50.0% ] .total cross section @xmath66 ( @xmath67 ) for accepted electron recoil events , ignoring ( including ) the effects of luminosity spread due to beamstrahlung .
the cuts applied are @xmath60 120 mrad and @xmath61 700 gev ( 350 gev for @xmath29 = 2 tev ) .
these are based on 20000 generated events each , so statistical errors are @xmath541% . [
cols="^,^,<,<",options="header " , ] 1.5 tev signal and background before ( left ) and after ( right ) applying the @xmath68 cuts .
the signal is represented by a black curve for two different values of @xmath69 , 0.4 for the higher peak and 0.1 for the lower .
the blue dotted line represents the @xmath70 background , green dotted is @xmath71 , and red dashed is their sum.,title="fig:",scaledwidth=50.0% ] 1.5 tev signal and background before ( left ) and after ( right ) applying the @xmath68 cuts .
the signal is represented by a black curve for two different values of @xmath69 , 0.4 for the higher peak and 0.1 for the lower .
the blue dotted line represents the @xmath70 background , green dotted is @xmath71 , and red dashed is their sum.,title="fig:",scaledwidth=50.0% ] the number of signal events passing the aforementioned cuts for @xmath29 = 1 , 1.5 , 2 tev and for a range of values of @xmath59 between 0.05 and 0.5 are plotted in figure [ fig : numevents ] with @xmath72 = 3 ab@xmath46 , and the corresponding confidence levels are plotted in figure [ fig : sigma ] .
the maximum value of @xmath59 at each mass has been chosen to keep the resonance width within reasonable limits .
it has been explained that we have not simulated the effects of beamstrahlung on the background .
the cross sections for both background processes diminish with increasing center of mass energy , so the effect of beamstrahlung would be to reduce the size of the background .
it can be seen in table [ tab:1 ] that the effect of beamstrahlung on the signal is to multiply it by a factor between @xmath56 0.5 and 0.65 , depending on the energy scale of the process .
the background has a similar energy dependence , so we would expect a roughly similar suppression of the background . although we can not be sure of the exact scale of the suppression , we can place an upper bound of 1.0 and a lower bound of 0.3 ( since @xmath56 30% of the luminosity is within 1% of the nominal center of mass energy ) , and take 0.65 as our expectation .
we represent these bounds by the shaded areas in figure [ fig : sigma ] . for each mass value
, the central line represents a 0.65 background suppression factor , the upper confidence bound represents a 0.3 suppression factor and the lower confidence bound represents no suppression ( factor 1.0 ) . although these bounds cover a broad range of confidence levels , we expect the true result not to differ very much from the central expectation value .
that pass the kinematic cuts with @xmath72 = 3 ab@xmath46 , with log scales on both the x and y axes .
the markers represent the data points .
green circles , blue squares and red triangles correspond to @xmath29 = 1.0 , 1.5 , 2.0 tev respectively .
each data point is calculated using 10000 signal events , with a statistical error of 1%.,scaledwidth=70.0% ] with @xmath72 = 3 ab@xmath46 from the signal @xmath73 .
the markers represent the data points . from top to bottom , the curves correspond to @xmath29 = 1.0 , 1.5 , 2.0 tev .
the filled regions represent the systematic error due to the unknown suppression of the background from the beamstrahlung luminosity spread , and are absolute bounds .
the upper and lower limits represent beamstrahlung background suppression factors ( see text ) of 0.3 and 1.0 respectively , and the central line 0.65 .
statistical errors are @xmath56 3%.,scaledwidth=70.0% ] the decay @xmath32 ( figure [ fig : feyn2 ] ) is allowed if @xmath75 .
for the small mixing angles considered here , it can be expected that the light state @xmath24 will have a mass close to that predicted for the sm , which has bounds 118 gev @xmath76 130 gev from direct searches @xcite , with another region opening up for @xmath77 , but this region is less favored by precision electroweak analysis .
this light higgs will decay with appreciable branching fraction into bottom quarks @xmath78 .
jets produced from @xmath78 quarks can leave a highly distinctive signature , and can be identified as such with some probability the @xmath78-tagging efficiency .
the channel @xmath79 therefore has the potential to give a very clear signal , as the backgrounds for @xmath80 are likely to be small at an electron - positron collider .
decay mode.,scaledwidth=60.0% ] a complete analysis of this would require varying four free observables
@xmath29 , @xmath81 , @xmath59 , @xmath82 , all of which contribute to the size of the signal , and it would be required to test that our points meet the existing theoretical and experimental constraints on the higgs sector . instead , we can use the data plotted in figure [ fig : numevents ] to estimate the number of @xmath80 events that we could expect to see , given some value of @xmath82 and @xmath83 .
if we require three or four @xmath78-tags , the probability of accepting an @xmath84 event is @xmath85 , where @xmath86 is the @xmath78-tagging efficiency .
a value @xmath87 @xcite gives @xmath88 , while a more conservative @xmath89 gives @xmath90 .
the number of @xmath80 events , @xmath91 , as a function of the number of @xmath92 events , @xmath93 , is then roughly @xmath94 having taken the values @xmath95 ( consistent with @xmath96 ) , @xmath97 , @xmath98 , @xmath99 .
so for instance , with @xmath87 and @xmath100 , we get @xmath101 , while @xmath89 would give @xmath102 .
clearly there are large regions of the parameter space that was explored in section [ subsec : zz ] that could also give a clear @xmath80 signal , though this is much more sensitive to the choice of parameters .
although current searches for new physics at the tev - scale are focussed on the experiments at the lhc , it should be remembered that there are possible states that could influence the electroweak symmetry breaking sector yet remain undiscovered there , but which could be discovered by new experiments that we currently have the capability of building .
one such possibility is the case where a hidden sector scalar mixes lightly with the sm higgs , resulting in two physical massive higgs particles .
the light mixing implies a heavily suppressed coupling to the sm for one of these particles , which makes its discovery very challenging .
in this paper we have investigated the prospects for finding a trans - tev higgs particle at a 3 tev electron - positron collider and have found that clic could substantially extend the reach for such searches compared to the lhc .
we have discussed three possible search channels : an electron recoil analysis , which would be particularly useful for an invisibly decaying higgs , the sm - like decay @xmath103 mode which is powerful and generic , and @xmath79 decays which could give a very clear and distinctive signal , shedding more light on the model parameters . data from a variety of channels could be combined to boost the confidence level for any discovery .
it must therefore be stressed that standard heavy higgs search channels that are normally considered for masses @xmath104 tev could be of importance well into tev energies , due to the possibility of strong width suppression . although the case for a multi - tev electron - positron collider depends on whether new physics is discovered at the lhc and on what form it takes , we believe that the potential for discovering new scalars that couple to a hidden sector , and which might not be found at the lhc , enhances that case .
j. kumar and j. d. wells , phys .
d * 74 * , 115017 ( 2006 ) [ hep - ph/0606183 ] .
p. langacker , rev .
phys . * 81 * , 1199 ( 2009 ) [ arxiv:0801.1345 [ hep - ph ] ] .
t. g. rizzo , hep - ph/0610104 .
atlas collaboration , observation of an excess of events in the search for the standard model higgs boson with the atlas detector at the lhc , " http://cdsweb.cern.ch/record/1460439 ( 9 july 2012 ) ; + cms collaboration , observation of a new boson with a mass near 125 gev , " http://cdsweb.cern.ch/record/1460438?ln=en ( 9 july 2012 ) .
r. schabinger and j. d. wells , phys .
d * 72 * , 093007 ( 2005 ) [ hep - ph/0509209 ] .
m. bowen , y. cui and j. d. wells , jhep * 0703 * , 036 ( 2007 ) [ hep - ph/0701035 ] . c. englert , t. plehn , d. zerwas and p. m. zerwas , phys .
b * 703 * , 298 ( 2011 ) [ arxiv:1106.3097 [ hep - ph ] ] . c. englert et al .
b * 707 * , 512 ( 2012 ) [ arxiv:1112.3007 [ hep - ph ] ] .
b. batell , s. gori and l. -t .
wang , arxiv:1112.5180 [ hep - ph ] .
l. linssen , a. miyamoto , m. stanitzki and h. weerts , `` physics and detectors at clic : clic conceptual design report , '' arxiv:1202.5940 [ physics.ins-det ] .
k. yokoya and p. chen , frontiers of particle beams : intensity limitations , eds .
m. month and s. turner , lecture notes in physics vol .
400 ( springer - verlag , berlin , 1990 ) , pp.414 - 445 . e. boos et al , [ comphep collaboration ] , ` _ _ comphep 4.4 : automatic computations from lagrangians to events _ _ ' , nucl .
a534 * ( 2004 ) 250 [ arxiv : hep - ph/0403113 ] ; + a.s .
belyaev et al , ` _ _ comphep - pythia interface : integrated package for the collision events generation based on exact matrix elements _ _ ' , proc . of acat2000 , fermilab , 16 - 20 october 2000 , p.211 [ arxiv : hep - ph/0101232 ] .
s. kortner , standard model scalar boson search with the atlas detector , " and m. pieri , searches for standard model
scalar boson at cms , " rencontres de moriond electroweak ( 7 march 2012 ) , http://indico.in2p3.fr/conferenceotherviews.py?view=standard&confid=6001 | the possibility of a scalar messenger that can couple the standard model ( sm ) to a hidden sector has been discussed in a variety of contexts in the literature in recent years .
we consider the case that a new scalar singlet charged under an exotic spontaneously broken abelian gauge symmetry mixes weakly with the sm higgs resulting in two scalar mass states , one of which has heavily suppressed couplings to the sm particles .
previous phenomenological studies have focussed on potential signatures for such a model at the large hadron collider ( lhc ) . however , there are interesting regions of the parameter space in which the heavier higgs state would be just out of reach for lhc searches if its mass is @xmath0 1 tev .
we therefore investigate the discovery potential for such a particle at a @xmath1 electron - positron collider , which is motivated by the recent r&d developments of the compact linear collider ( clic ) .
we find that such an experiment could substantially extend our discovery reach for a heavy , weakly coupled higgs boson , and we discuss three possible search channels . |
two decades ago , @xcite demonstrated through numerical models that gravitational interactions between galaxies can generate spectacular tidal features . in earlier work , @xcite already proposed that collisions between giant galaxies may result in the ejection of stars and gas into the intergalactic space , resulting in the formation of star clusters and star forming dwarf galaxies .
direct observational evidence for the existence of such a class of recycled galaxies has already been reported in the literature ( e.g. * ? ? ?
systems like the antennae @xcite and the super antennae
@xcite show , in addition to stellar material and gas torn from the disks of the interacting galaxies , optically bright clumps along and at the extremes of the tidal tails as well . in some cases
there is even a chain of knots of bright material whose colors become bluer towards the tips @xcite .
radio observations of these systems have shown that a large amount of atomic hydrogen is torn away from the parent galaxies , stretched along the tidal tails , and seems to concentrate at the tips .
in contrast , molecular hydrogen is concentrated in the central regions of the parent galaxies .
a particularly nice example is the interacting galaxy arp105
@xcite , that shows near the end of its tidal tail a lmc like galaxy , resulting from the collision of a spiral with an elliptical galaxy .
it is well known from numerical simulations that the form and strength of the tidal bridges and tails depend on the orbital and spin parameters of the colliding galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , the star formation activity within the tidal tails could only recently be numerically studied , after the development of appropriate codes ( e.g. * ? ? ?
* ; * ? ? ?
models that do not take into account the gas , or the fact that the is distributed predominantly at the periphery of spirals , are unable to reproduce the observed distribution after a collision @xcite , while those that do take this into account find that , after a collision with a massive body , the gas can form huge pools at the extreme ends of the tidal tails @xcite .
in this paper we study the group ngc6845 ( klemola30 , am1957 - 471 ) .
@xcite described the group as being compact , dominated by a bright , one armed spiral . in the @xcite catalogue , ngc6845
is described as an interacting quartet .
its four components are approximately aligned in a sw ne direction .
the components ngc6845a and ngc6845b , the later located to the ne , show abundant signs of an ongoing interaction , such as tidal bridges and tails .
many clearly defined knots dot the tidal bridge that links ngc6845a to ngc6845b .
signs of interaction seem also to be present in ngc6845c and ngc6845d , according to @xcite and @xcite sketches . in order to study the collision kinematics and the subsequent process of star formation
, we carried out optical imaging and spectroscopy as well as radio synthesis observations . in section
[ sec : obs ] we show our optical and radio observations of ngc6845 and briefly describe the data reduction ; in sec .
[ sec : res ] we present our analysis ; two possible scenarios for the interaction process are discussed in sec . [
sec : dynamics ] and in sec .
[ sec : conclu ] we summarize our results and present our conclusions .
table [ tab : gal ] presents some basic data on ngc6845 .
our observations consist of optical broadband photometry and spectroscopy obtained at the casleo and bosque alegre observatories and observations taken with the very large array .
tables [ tab : ima ] and [ tab : spec ] list the relevant instrumental and observational parameters .
we adopted h@xmath2 thoughout this paper , which results in a distance to ngc6845 of 85mpc . at this distance , 1
arcsec corresponds to 412pc .
the observations are described in more detail below .
broad band b , v , r and i ccd images were obtained in september 1997 with the ritchey
chrtien 2.15m telescope at casleo , argentina , using a thinned tektronix @xmath7 pixel ccd detector .
the telescope focal ratio is f/8 .
the final image scale , with a focal reducer , is @xmath8 .
the seeing during the observations varied between 2 and 45 .
data reduction was performed with iraf , using standard procedures .
different frames taken through the same filter were registered to a common center , using field stars as a reference , and co added to remove cosmic ray events and increase the dynamic range .
small seeing differences among frames were corrected by gaussian filtering .
flux calibration was based on standard stars from @xcite .
we corrected as usual for atmospheric extinction .
long slit spectroscopic observations were carried out on july 1994 ( seeing=17 ) and september 1997 ( seeing=2 ) with the 2.15 m telescope of the casleo observatory , and on december 1997 with the bosque alegre 1.52 m telescope ( seeing=15 ) . in table [ tab : spec ] we present the log of the spectroscopic observations .
the slit positions , given in the first column of this table , are marked in figure [ fig : slit ] .
data processing was again done with iraf following standard methods for bias subtraction , flat fielding , and image combination .
the resulting two dimensional spectra were wavelength calibrated using comparison spectra and flux calibrated using spectrophotometric standard stars from @xcite .
galactic extinction was corrected with the standard absorption law , adopting @xmath9 @xcite .
array observations of the ngc6845 system were obtained during a short , 1.5 hour run in 1996 , june 19 . due to its far southern declination the synthesized beam
is substantially elongated in the n
s direction .
the beam measures @xmath10 , at a @xmath11 angle measured from north to east .
the sensitivity of the system was rather modest and the data should be seen as exploratory . as it was not obvious at what velocities
could be expected we aimed for velocity coverage rather than for velocity resolution , trying to maximize our sensitivity for the velocity range expected for objects a and b ( radial velocities are given in tab . [ tab : gal ] ) .
the velocity resolution was @xmath12 , covering the interval @xmath13 to @xmath14 with both right hand ( r ) and left hand ( l ) polarization ; the intervals from @xmath15 to @xmath13 and @xmath14 to @xmath16 , were observed with one polarization only .
data calibration and mapping was performed with aips .
the final maps have an rms noise of @xmath17 or 0.23kelvin ( where @xmath18 corresponds to 0.15k ) .
we present in figure [ fig : rimag ] the r image , enhancing the lowest surface brightness levels in order to highlight the faint details that might have been produced by the interaction .
clear indications of strong collision are observed in ngc6845a and ngc6845b .
the tidal bridge between galaxies a and b present a fainter counterpart , a tidal tail that bifurcate and seems to connect to components c and d ( see sketches in @xcite and @xcite and plate i in @xcite ) . in table
[ tab : phot_gal ] we present the integrated b magnitude and colors ( b - v ) , ( v - r ) and ( v - i ) of the group members . as can be seen from tab .
[ tab : gal ] , our b magnitudes are in better agreement with @xcite than with @xcite .
the ( b - i ) color map ( fig .
[ fig : colorbi ] ) shows that the color of the tidal bridge is bluer than the inner disk of a and b components and comparable to their outer disks .
@xcite pointed out that in the case of the evolved merge ngc7252 the tidal tails are 0.20.5 mag bluer in ( b - r ) than the outer regions of the main body .
ngc6845a is classified as sb(s)b ( @xcite ) .
the color distribution along its bar is not uniform . to the sw ,
the bar is much redder , indicating that on this side we are seeing dust in front of the stellar bar , where it is usually accumulated in more or less linear shock fronts . the bar of ngc6845a is shifted by @xmath19 counterclockwise with respect to the photometric major axis , and it may have been tidally created by the encounter , as suggested for example by @xcite or @xcite s @xcite simulations of disk - disk encounters .
plumes and small condensations are evident around ngc6845b , and are likely debris of the tidal interaction .
the tidal bridge linking the ngc6845a and b galaxies is very pronounced and knotty ( fig .
[ fig : zoomvel ] ) .
the knots are regions , as will be shown in sec .
[ subsec : knots ] .
@xcite s ( 1979 ) ctio 4 m plate shows more clearly than our fig .
[ fig : rimag ] the presence of condensations in the tidal tail , specially a set of them close to the nw side of ngc6845c , though it does not shows the bridge knotty structure because it is over exposed .
based on @xcite s plate , @xcite schematically suggested that ngc6845a western tidal tail bifurcates and links to ngc6845c and d galaxies .
table [ tab : phot_reg ] shows the v magnitude and ( b - v ) , ( v - r ) and ( v - i ) color indexes for the most conspicuous knots marked in fig [ fig : zoomvel ] . without exception
the knots present an region like spectrum , as shown in figure [ fig : spectra ] , and we feel therefore confident identifying them as regions of recent star formation . in order to correct the colors for internal absorption we used for all regions the mean value of the @xmath20 ratio as was measured in knots 4 , 7 and 8 ( see tab .
[ tab : flux ] ) . and
intensities were measured from slit positions 3 and 5 for knot 4 and slit position 4 for knots 7 and 8 ( see fig .
[ fig : slit ] ) .
although these regions are located at very different positions within the system , they present roughly the same @xmath20 ratios , so we assume a constant @xmath20 ratio of 4.3 to correct the other regions through the formula : @xmath21 @xcite .
we compared the brightness and colors of the knots to the evolutionary synthesis models of @xcite , with updatings ( girardi 1999 , private communication ) .
we chose from their models those with a salpeter imf and a heavy element abundance @xmath22 . in table
[ tab : phot_reg ] we present the mean age of the knots obtained by fitting our three colors to girardi s models .
their masses were derived from the v magnitude and assuming a distance of 85mpc . within the errors
all the knots , except for numbers 6 and 8 , are coeval to within our stated accuracy .
the derived values , quoted in table [ tab : phot_reg ] , imply that we are dealing with star clusters with masses@xmath235@xmath24m@xmath25 . in table
[ tab : flux ] we present the emission line luminosities of all the knots in fig .
[ fig : zoomvel ] , plus the nuclear emission of ngc6845a and ngc6845b .
as can be seen , knots 1 , 2 and 3 , located in the middle of the tidal bridge , have luminosities comparable to that of ngc3603 , the most luminous region known to exist in our galaxy ( l@xmath26 , @xcite ) or higher . regions 4 and 5 , at the tip of the tidal bridge are even brighter , being similar to 30doradus ( l@xmath27 , @xcite ) .
giant regions near the ends of tidal bridges and tails were reported for various systems , as in ngc4676 by @xcite and in ngc2623 , ngc6621/22 , ngc3256 and ngc2535/36 by @xcite .
these ionising clusters are usually more luminous than the ones along the tidal tails . finally , regions 7 and 8 , located at the extreme ends of the bar in ngc6845a , have luminosities comparable to the `` jumbo '' region of ngc3310 ( l@xmath28 @xcite ) , indicating a strong star formation activity in these places , which may also have been triggered by the collision @xcite . like other interacting systems ,
this group is prominent in the infrared .
the iras fluxes ( f@xmath29jy , f@xmath30jy , f@xmath31jy and f@xmath32jy ; @xcite ) , lead to a group total luminosity of @xmath33 , locating this group at the fainter end of luminous ir galaxies according to the definition by @xcite . in fig .
[ fig : r_hi ] we show the isodensity contours superimposed on the r band image . despite the low spatial resolution of the radio observations , it is possible to distinguish two velocity components that differs by a few hundred @xmath34 , a main component centered on ngc6845a and a secondary one associated with ngc6845b .
no was detected at the velocities of galaxies c and d. the main component presents a clear extension towards the sw , bending south , following the stellar tidal tail . to the east , it shows a protrusion that coincides with the bright optical tidal bridge , with velocities compatible with this feature within the observational errors .
the integrated flux of the main component is @xmath35 , which translates to an mass of @xmath36 .
the component associated with ngc6845b is virtually stationary in velocity .
the mean velocity is around 6700kms@xmath37 , close to that of ngc6845b .
spatially , it is slightly shifted eastward with respect to ngc6845b .
the integrated flux is @xmath38 , which translates to @xmath39
. the total integrated flux of both component is @xmath40 , corresponding to an mass of @xmath41 , or about 5 times more than that contained in the milky way . in fig .
[ fig : slit ] we show the slit positions along which we obtained spectra . for ease of reference we indicated in fig .
[ fig : zoomvel ] the radial velocity of the more relevant objects .
radial velocities were obtained using a weighted mean of the values measured for the , and lines .
a comparison with table [ tab : gal ] shows that the radial velocity values we measured for ngc6845a and b are in agreement with previous results from @xcite and @xcite .
there is , however , a discrepancy between the values given by the cited references for radial velocities of components c and d. apparently one of this sources made a mistake in the galaxies identification .
the rotation curve of ngc6845a , presented in figure [ fig : rota ] , was obtained from slit position 4 ( see table [ tab : spec ] and fig . [
fig : slit ] ) , along the apparent major axis of the galaxy .
line of sight velocity measurements were transformed to circular velocities adopting an inclination angle of @xmath42 , derived by fitting ellipses to the optical isophotes of the b image ( using the appropriated procedure in the stsdas
iraf package ) . since the isophotes in the inner part of ngc6845a are influenced by the bar profile , and those more external are tidally distorted , we adopted the ellipticity of isophotes at intermediate radii as the best indicator for the inclination .
there is a marked difference between the ne and the sw sides of the rotation curve for r@xmath436kpc ( see fig .
[ fig : rota ] ) : to the ne it shows a strong bump at about 6kpc , whereas the sw half presents a plateau at about the same distance .
a qualitative comparison with numerical simulations ( e.g. * ? ? ?
* ; * ? ? ?
* ) suggests that the difference is more likely due to tidal rather than intrinsic effects , which perturb more strongly the galactic disk periphery . the average rotation curve ( fig . [ fig : rota ] )
was fitted with @xcite s ( 1980 ) density law : @xmath44^{-3/2}\ ] ] where @xmath45 is the circular velocity at position @xmath46 , m is the total mass and ( a , b ) are shape parameters with the dimension of length .
following satoh , a ratio of @xmath47 corresponds to a disk dominated distribution , @xmath48 to a disk plus a well developed bulge ( sa so , see * ? ? ?
* ) and @xmath49 to a near spherical distribution .
@xmath50 fitting was performed by the downhill simplex minimization algorithm ( `` amoeba '' ) with the iraf nfit1d task .
the parameters obtained for the fit were @xmath51 and @xmath52 , which indicates a dominant disk with a small to intermediate bulge , in agreement with @xcite classification the fit is shown in fig .
[ fig : rota ] , where the full line represents the best fit and the dashed lines give its lower and upper limits within the observational errors , which in turn furnishes @xmath53 and @xmath54 errors .
the derived model total mass is @xmath55 .
the keplerian mass up to the last point on the rotation curve is @xmath56 , broadly in agreement with the more sophisticated fit based on @xcite s model .
ngc6845a mass is comparable to that of andromeda .
andromeda s brightness is @xmath57mag at a distance of 0.7mpc , corresponding to 13.5mag at the distance of ngc6845a , and indicating that both galaxies are comparable in their m / l ratio . nevertheless , the size of ngc6845a is only half that of andromeda
. this figure would be difficult to change since in order to make both galaxies comparable in size , a hubble constant smaller than @xmath58 would be required .
it is also difficult to invoke an error in the derived mass for ngc6845a due to an error in the disk inclination since , even by assuming an edge - on position for this galaxy , we could not reduce the derived mass by more than a factor of 1.5 .
@xcite pointed already out that ngc6845a should have a large mass because of its uncommonly large maximum velocity .
we note that their rotation curve extends only up to @xmath59 .
it is interesting to note that the tidal bridge of ngc6845a shows a decreasing trend in radial velocity ( see fig [ fig : zoomvel ] and tab [ tab : flux ] ) starting from its base , where it is connected to the ngc6845a disk near region 8 ( at 6660@xmath60 ) , and passing through regions 3 , 2 and 1 , located more or less in the middle of the tidal bridge ( at 6455@xmath60 , 6361@xmath60 and 6272@xmath60 , respectively ) .
regions 4 and 5 , which are slightly shifted to the north with respect to the prolongation of the bridge , surprisingly present radial velocities that agree with this trend ( 6314@xmath60 and 6201@xmath60 , respectively ) , strongly suggesting that they are linked to the tidal bridge and not to ngc6845b , despite the fact that they are seen projected on top of this object .
we also detected on the tidal bridge , with a mean velocity of 6350@xmath60 , in agreement with the optical results ( see next section ) .
supposing that the tidal bridge tip follows a parabolic orbit in the plane of ngc6845a disk and its present true velocity is the observed radial velocity of the regions 4 and 5 , we obtain a collision age of @xmath61150300myr . since regions can not be older than 10myr ( see sec .
[ subsec : knots ] ) , this implies that star formation activity occurs inside the tidal bridge , started well after the beginning of the bridge formation .
a possible scenario for the bridge late star formation come out from @xcite simulations .
these authors found that self gravitating star clumps forms along tidal tails after a collision .
the gas clouds torn apart from the galactic disk simultaneously with the stars may fall into these clumps , reaching the necessary density in order to form stars .
cloud collisions inside the tidal bridge may be an alternative mechanism to explain the star formation inside the bridge .
the isovelocity map of the component associated with ngc6845a is shown in fig .
[ fig : r_vel ] , superimposed on the r band image . in fig .
[ fig : prof ] , we show the position
velocity diagram , obtained by cutting the data cube along the major axis of ngc6845a ( line of the optical spectrum slit number 4 in figure [ fig : slit ] ) . in this figure
the center of ngc6845a is at @xmath62 and @xmath63 . to the south of the galaxy center , the diagram shows a behavior similar to that of the optical rotation curve ( fig .
[ fig : rota ] ) , so we conclude that in this part the kinematics is mainly rotational . to the north ,
the lower part of this diagram shows a plateau at @xmath64 , coincident with the tidal bridge velocity .
supposing that the is in dynamical equilibrium , we derive from its kinematics ( fig . [
fig : r_vel ] ) a mass of 1.4@xmath65 , about three times higher than the mass derived from the optical data .
we attribute this difference to the fact that the can be followed out to larger distances , especially to the sw ( the upper part of the position velocity diagram of fig .
[ fig : prof ] ) , and hence will trace the dark matter associated with the halo of ngc6845a
. nevertheless , we have to keep in mind that the kinematics may suffer non rotational effects .
the map low spatial resolution does not allow to see the classical butterfly diagram , characteristic of 2-d rotational maps .
as can be seen in figure [ fig : r_vel ] , the isovelocity lines are roughly oriented along the beam major axis . following @xcite , equation 1021 ) , the radial velocity difference among the system members , their absolute luminosities ( tab .
[ tab : phot_gal ] ) and projected positions , allow us to derive the group m / l ratio .
for the group member velocities we adopt the mean value of the available data ( @xcite ; @xcite and ours , see tab .
[ tab : phot_gal ] ) .
an important source of error in determining
m / l is the high inclination of the c component and , to a less extent , that of the a component , the group most massive galaxies .
the inclination of each group member was obtained as quoted in sec .
[ subsubsec : optkin ] for ngc6845a .
@xcite s ( 1958 ) correction was applied to the measured isophotal axis ratio @xmath54 by the formula cos@xmath66 in order to obtain the galaxies inclinations ( see tab . [
tab : phot_gal ] ) , where @xmath67 @xcite .
this correction is specially important for ngc6845c .
we obtained for this galaxy @xmath68 , measured at the isophotal level 25 mag / arcsec@xmath69 .
galactic extinction was corrected with the standard absorption law @xcite . in order to obtain the galaxies
face on magnitudes we corrected internal absorptions by the empirical ( @xcite ) formula @xmath70 , which was evaluated following two different prescriptions of the values of @xmath71 and @xmath72 .
the first one is from @xcite , where @xmath73 and @xmath74 , leading to ( m / l)@xmath75 ( solar units ) , with the uncertainties estimated from the errors in @xmath72 @xmath76 , @xcite and @xmath77 ( tab .
[ tab : phot_gal ] ) .
the second prescription is that of @xcite , where @xmath78 and @xmath79 for ngc6845a and @xmath80 and @xmath81 for galaxies b , c and d. in this case we obtained ( m / l)@xmath82 .
m / l in the range 2849 , also depending on the source for galactic and internal extinction corrections . our ( m / l)@xmath83 is well inside their range of values , while ( m / l)@xmath84 is higher than their highest m / l ratio .
although large , the difference between ( m / l)@xmath83 and ( m / l)@xmath84 points to a very similar total mass of the system ( @xmath85 and @xmath86 respectively ) .
the luminous mass of the system can not account for more than @xmath87 ( scaling the mass of the galaxies by their luminosities with respect to the mass and luminosity of ngc6845a ) , indicating that the group s dark matter amounts to approximately 5 times higher than its luminous mass .
the whole system contains almost 9 times more dark matter than what was thought to be associated with the halo of ngc6845a , as derived in the previous section from kinematics .
strong , well developed tidal bridges and tails are characteristic of galaxy encounters in which the disk of the perturbed galaxy and the orbit of the perturber rotate in the same sense , i.e. , a prograde orbit , and the angle between the main galaxy disk and the orbital plane is small ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in prograde encounters , tidal arms curve in a sense opposite to the disk rotation
: they are trailing .
consequently , we can infer that ngc6845a disk rotates counterclockwise ( sec .
[ subsubsec : optkin ] ) and its se side is preceding , indicating that the tidal bridge side is the near one .
@xcite already noted as curious that due to the relative position and radial velocities of galaxies a and b , any simple geometry would indicate that the b orbit is retrograde .
however , as mentioned by these authors , the case might be similar to that of m51/ngc5195 , for which @xcite have found it possible to obtain a direct sense of rotation if the orbital eccentricity of ngc5195 is large and if the orbit is highly inclined to the plane of m51 . from this point of view
, galaxy c would be a good candidate for the strong interaction .
nevertheless , this galaxy presents itself as a symmetric disk up to faint brightness levels .
modern numerical simulations of similar disk encounters ( e.g. * ? ? ?
* ) , shows that a collision between two large disks , like those of ngc6845a and c , close enough as to produce ngc6845a tidal bridge , would necessarily strongly distort ngc6845c disk as well . in order to understand the dynamical evolution of this group , two of us (
* , in preparation ) , are developing extensive numerical simulations with the treesph code @xcite in a cray t94 computer ( cesup ufrgs ) . from this work we advance some preliminary results . in figs .
[ fig : simulac ] and [ fig : simulab ] , we shows simulations of a with c and a with b collisions .
galaxy models were constructed following the general prescription of @xcite .
galaxy a model is such that its rotation curve presents a circular velocity of 300kms@xmath37 at 6kpc , and is approximately flat up to the cutoff radius ( 20kpc ) .
10% of the disk mass is composed of gas particles , whose density profile is more extended than that of the stellar disk . for galaxy c
we adopted the same model of a , but without gas , since this galaxy do not present or emission lines in its optical spectrum .
a and c galaxy models consist of 6000 particles with a mass of 3.5 @xmath88 per galaxy .
galaxy b is represented by a point mass of 1/4 of model a mass .
[ fig : simulac ] shows the evolution of the encounter of models a and c , which approach along a parabolic orbit with a perigalactic distance @xmath89kpc .
the first row shows the time evolution of the stellar component , while the second row shows the same for the gas component . for easy of comparison between simulation and observations we used a coordinate system in which the z axis corresponds to the line of sight ( with positive values towards the observer ) , the x axis is the e - w direction ( positive to the west ) and the x axis is the n - s direction ( positive to the north ) . in this system
, the orbit of c around a has its pericenter in the direction of the vector @xmath90 , and the orbital plane is defined by the orbital spin vector , parallel to @xmath91 .
we identify the third frame of fig .
[ fig : simulab ] ( 104myr after the perigalacticum ) as the more similar to the system present stage .
a striking result in favor of this scenario is the presence of a stellar condensation similar to that observed to the right of galaxy c nw extreme .
c model se extreme bend to the e , which is also observed to the faintest brightness levels in galaxy c ( fig .
[ fig : rimag ] ) .
the simulation also clearly shows that this collision produces a strong tidal feature similar to the ngc6845a tidal bridge with a condensation near its extreme that resembles regions 4 and 5 , but nothing as large as galaxy b is produced out of this encounter , so the position of galaxy b at the tidal bridge extreme would be a projection within this scenario . against this scenario
the simulation shows that about 10% of the gas originally belonging to galaxy a would be captured by galaxy c , which is neither observed as nor as .
the simulation also shows that stars stripped out from a should form a stellar feature perpendicular to c disk , also not detected in the optical images , which present a high degree of symmetry to intermediate brightness levels .
[ fig : simulab ] shows the encounter of model a with the point mass b , under a prograde hyperbolic orbit with eccentricity @xmath92 and perigalactic distance @xmath93kpc .
the pericenter is in the direction @xmath94 and its orbital spin vector is parallel to @xmath95 .
the third frame ( 145myr after the perigalacticum , shown in the second frame ) shows the best match of the model to the system present stage . in this frame galaxy b
is receding with respect to a at @xmath96 .
the more striking feature of this simulation is the morphology of the tidal bridge , which is straight and narrow and shows condensations of gas and stars at its base and middle part , resembling the observed tidal bridge morphology .
also the tidal tail is diffuse , as in the real situation . against this scenario
the simulation shows that the tidal tail points in a direction different from the observed one and does not bifurcate . in spite of both scenarios pros and cons
, the collision of a and b seems to be more appropriate to explain the tidal bridge structure .
although some features associated to galaxy c can be regarded as tidally produced , it can not be ignored the lack of other important features that c should show if it were the main perturber .
perhaps a three body interaction , with b producing the tidal bridge would be more suitable to explain the richness of features observed in this group .
the system ngc6845 is a compact group in which ngc6845a presents a strong tidal bridge , signature of a close collision .
we present the first atomic hydrogen observation of ngc6845 , obtaining an content 5 times higher than in the case of the milky way .
the distribution can be separated into two components , one related to ngc6845a , sporting well developed tidal features and the second one , with velocity similar to that of ngc6845b .
components c and d do not present , up to our detection limit .
this fact set constrains to galaxy c being the main perturber of a , since , as shown by numerical simulations , in these case gas originally belonging to a should be transferred to c. ngc6845a optical rotation curve is asymmetric , with a uncommonly high maximum circular velocity , especially to the sw side .
satoh model fitting to the rotation curve indicates the presence of a dominant disk and a small to intermediate bulge , in agreement with the morphological classification of this galaxy .
the derived mass is m@xmath97 . the ( b - i ) color map indicate the presence of dust in front of the sw side of the bar , pointing to a trailing bar .
the ( b - i ) color map also shows that the bridge is bluer than ngc6845a inner disk .
regions brighter than those of the antennae and ngc7252 tidal tails have formed at the tip and at the base of the tidal bridge .
three regions midway along the tidal bridge are brighter than ngc3603 , the brightest one known in our galaxy .
the kinematics of regions 4 and 5 indicates that they are more likely connected to the tidal bridge than to ngc6845b , as their projected position may suggest .
they present luminosities similar to that of 30doradus .
a simple model for the kinematics of the tip of the bridge as well as numerical simulations indicate that star formation is occurring along this structure one hundred million years or more after the perturber closest approach .
the group kinematics calls for dark matter within the system at about 5 times its luminous mass .
the group amount of dark matter would be 9 times higher than the halo mass of the ngc6845a , provided that the is in dynamical equilibrium inside the galaxy ; this in turn would be three times higher than the galaxy kinematical mass derived from the optical rotation curve .
since abundant signs of interactions are present in galaxy b , and to a less extent in c , we appealed to n - body simulations in order to disentangle which of the galaxies , b or c , would be the a main perturber .
both scenarios for the tidal disruption of galaxy a present pros and cons , but the bridge , which is the most conspicuous tidal feature of this galaxy , seems to be better reproduced by the simulation of the encounter of galaxies a and b. the fact that neither nor , that should have been stripped out from galaxy a , are detected in galaxy c reinforce this scenario . perhaps a three body interaction would be a more appropriate scenario to explain the group dynamical evolution , but it is difficult to model due to the large number of degrees of freedom involved such a collision .
we are grateful to dr .
lars hernquist , who kindly made available the treesph code , to mr .
ruben diaz , who carried out the observations at bosque alegre and to dr .
leo girardi , who kindly provided his evolutionary synthesis models .
the authors greatefully acknowledge the nrao for allocating observing time and for support .
we also thank the staff of casleo and bosque alegre .
numerical calculations were done at the cesup - ufrgs national supercomputing center .
this work was supported in part by the brazilian institutions cnpq and finep .
arp , h.c . , and madore , b.f .
1987 , a catalogue of southern peculiar galaxies and associations , cambridge university press barnes j. 1988 , , 331 , 699 barnes j. 1992 , , 393 , 484 barnes , j. , and hernquist , l. 1992 , , 360 , 715 barton , e.j . ,
bromley , b.c .
geller , m.j .
1999 , , 511 , l25 bertelli , g. , bressan , a. , chiosi , c. , fagotto , f. , nasi .
e. 1994 , , 106 , 275 binney , j. , and tremaine , s. 1987 , _ galactic dynamics _ ( princeton series in astrophysics ) burstein , d. , and heiles , c. 1984 , , 54 , 33 de vaucouleurs , g. , de vaucouleurs , a. , corwin jr .
buta , r. j. paturel , g. , and fouqu , p. 1991 , _ third reference catalogue of bright galaxies _
( springer - verlag , new york)(rc3 ) duc , p.a , and mirabel , i.f .
1994 , , 289 , 83 duc , p.a , brinks , e. , wink , j. , & mirabel , i.f .
1997 , , 326 , 537 elmegreen , b.g . ,
kaufmann , thomasson , m. 1993 , , 412 , 90 eneev , t.m . ,
kozlov , n.n . , and sunyaev , r.a .
1973 , , 22,41 graham , j.a . , and rubin , v.c .
1973 , , 183 , 19 hernquist , l. , katz , n. 1989 , , 70 , 419 hibbard , j.e . ,
guhathakurta , p. , van gorkon , j.h . , schweizer , e. 1994 , , 107 , 67 holmberg , j.h .
1958 , med .
lunds astron .
ii , no . 136 howard , s. , keel , w.c . , byrd , g. , burkey , j. 1993 , , 417 , 502 kennicutt , r.c . , and chu , y.h .
1988 , , 95 , 720 kennicutt , r.c . , edgar , b.k . , and hodge , p.w .
1989 , , 337 , 761 klemola , a.r .
1969 , , 74 , 804 kuijken , k , and dubinski , j. 1995 , , 277 , 1341 kodaira , k. , watanabe , m. 1988 , , 96 , 1593 landolt , a.u .
1992 , , 104 , 372 melnick , j. , tapia , m. , and terlevich , r. 1989 , , 213 , 89 mihos , j.c . , and bothun , g.d .
1997 , , 481 , 471 mirabel , i.f . ,
dottori , h. , and lutz , d. 1992 , , 256 , l19 mirabel , i.f . ,
lutz , d. , & maza , j. 1991 , , 243 , 367 negroponte , j. , and white , s.d.m .
1983 , , 205 , 1009 osmer , p.s . ,
smith , m.g . , and weedman , d.w .
1974 , , 192 , 279 pastoriza , m.g . ,
dottori , h.a . ,
terlevich , e. , terlevich , r. , diaz , a. 1993 , , 260 , 177 rodrigues , i , and dottori , h. 1999 , _ in prep .
_ rose , j.a .
1979 , , 231 , 10 rose , j.a . , graham , j.a .
1979 , , 231 , 320 rubin , v.c . ,
ford , w.k . , jr . ,
thonnard , n. , and burstein , d. 1982 , , 261 , 439 rush , b. , malkan , m.az . , and spinoglio , l. 1993 , , 89 , 1@xmath71 and @xmath72 salo , h. 1991 , , 243 , 118 satoh , c. 1980 , , 32 , 41 sanders , d.b . , and mirabel , i.f .
1996 , , 34 , 749 schweizer , f. 1978 , iau symp .
77 , 279 stockton , a. 1974 , , 187 , 219 stone , r.p.s . , and baldwin , j.a .
1983 , , 204 , 347 toomre , a. , and toomre , j. 1972 , , 178 , 623 wright , a.e .
1972 , , 157 , 309 zwicky , f. 1956 , erg .
bd , 29 , 344 l c c c c l @xmath72(1950 ) & @xmath98 & @xmath99 & @xmath100 & @xmath101 & @xcite + @xmath102(1950 ) & @xmath103 & @xmath104 & @xmath105 & @xmath106 & @xcite + l & 352.24 & 352.26 & 352.23 & 352.21 & @xcite + b & -30.99 & -31.02 & -30.99 & -30.98 & @xcite + @xmath107 ( mag ) & @xmath108 & @xmath109 & @xmath110 & @xmath111 & @xcite + @xmath112 ( mag ) & @xmath113 & @xmath114 & @xmath115 & @xmath116 & @xcite + @xmath107 ( mag ) & @xmath117 & @xmath118 & @xmath119 & @xmath120 & this paper ( see tab .
[ tab : phot_gal ] ) + @xmath121 & @xmath122 & @xmath123 & @xmath124 & @xmath125 & @xcite + @xmath121 & @xmath126 & @xmath127 & @xmath128 & @xmath129 & @xcite + @xmath121 & @xmath130 & @xmath131 & & & this paper ( see tab . [ tab : flux ] ) c c c c c c r c c 1 & jul 08 1994 & b & c & 300 & 4100 - 6900 & 8 & 4.5@xmath132170 & 3 & 4800 + 2 & jul 08 1994 & b & c & 300 & 4100 - 6900 & 8 &
6@xmath132170 & 3 & 10800 + 2 & sep 08 1997 & reosc & 600 & 5870 - 7460 & 5 & 3.6@xmath132120 & 1 & 3600 + 2@xmath133 & dec 18 1997 & multi - function & 1200 & 6130 - 6850 & 2 & 1@xmath132130 & 1 & 3600 + 3 & jul 09 1994 & b & c & 300 & 4100 - 6900 & 10 & 2@xmath132170 & 3 & 3200 + 4 & jul 10 1994 & reosc & 300 & 3700 - 7150 & 8 & 2@xmath132120 & 2 & 1800 + 5 & jul 10 1994 & reosc & 300 & 3700 - 7150 & 8 & 6@xmath132120 & 2 & 2700 + 6 & sep 08 1997 & reosc & 600 & 5870 - 7460 & 6 & 2.3@xmath132120 & 4 & 9300 + 7 & sep 08 1997 & reosc & 600 & 5870 - 7460 & 6 & 2.3@xmath132120 & 1 & 3600 a & 13.84 & 0.64 & 0.23 & 0.20 & 0.54 & 0.16 & 0.50 & 0.32 & -21.56 & -21.38 & 3.2 & 2.7 + & @xmath1340.02 & @xmath1340.03 & @xmath1340.07 & @xmath1340.14 & @xmath1340.06 & & & & & & @xmath135 & @xmath136 + b & 15.19 & 0.64 & 0.26 & 0.29 & 1 & 0.16 & 0.00 & 0.10 & -19.71 & -19.81 & 5.9 & 6.5 + & @xmath1340.02 & @xmath1340.03 & @xmath1340.10 & @xmath1340.22 & @xmath1340.08 & & & & & & @xmath137 & @xmath138 + c & 14.35 & 0.81 & 0.35 & 0.16 & 0.22 & 0.16 & 1.86 & 0.82 & -22.41 & -21.37 & 7.1 & 2.7 + & @xmath1340.02 & @xmath1340.04 & @xmath1340.10 & @xmath1340.17 & @xmath1340.05 & & & & & & @xmath135 & @xmath139 + d & 15.95 & 0.83 & 0.35 & 0.11 & 0.33 & 0.16 & 1.06 & 0.50 & -20.01 & -19.45 & 7.8 & 4.6 + & @xmath1340.02 & @xmath1340.04 & @xmath1340.08 & @xmath1340.18 & @xmath1340.03 & & & & & & @xmath137 & @xmath138 + c c c r r c c 1 & [email protected] & [email protected] & [email protected] & [email protected] & @xmath140 & @xmath141 + 2 & [email protected] & [email protected] & [email protected] & [email protected] & @xmath142 & @xmath143 + 3 & [email protected] & [email protected] & [email protected] & [email protected] & @xmath144 & @xmath145 + 4 & [email protected] & [email protected] & [email protected] & [email protected] & @xmath146 & @xmath147 + 5 & [email protected] & [email protected] & [email protected] & bn [email protected] & @xmath146 & @xmath148 + 6 & [email protected] & [email protected] & [email protected] & [email protected] & @xmath149 & @xmath150 + 7 & [email protected] & [email protected] & [email protected] & [email protected] & @xmath151 & @xmath152 + 8 & [email protected] & [email protected] & [email protected] & [email protected] & @xmath153 & @xmath154 c c c c c c 1 & @xmath155 & & @xmath156 & 2 & 6272 + 2 & @xmath157 & & @xmath156 & 2 & 6361 + 3 & @xmath158 & & @xmath156 & 2 & 6455 + 4 & @xmath159 & 4.2 & @xmath160 & 7;3;5 & 6314 + 5 & @xmath161 & & @xmath160 & 6 & 6201 + 6 & & & & & + 7 & @xmath162 & 4.1 & @xmath163 & 4 & 6113 + 8 & @xmath164 & 4.6 & @xmath165 & 4 & 6660 + ngc6845a & @xmath166 & & @xmath167 & 4 & 6399 + ngc6845b & @xmath168 & & @xmath169 & 6;7 & 6746 + | we present optical spectroscopy , b , v , r and i ccd photometry and vla neutral hydrogen observations of the interacting quartet ngc6845 , also know as klemola30 .
ngc6845a , the dominant component , sports a broad and bright tidal bridge and a faint tidal tail , which bifurcate .
the tidal bridge has a ( b - i ) color bluer than that of ngc6845a inner disk .
five strong condensations , identified as regions brighter than the brightest one known in our galaxy , are found along the tidal bridge , with the two most luminous located at the bridge tip .
two giant regions , comparable to 30dor , are located where the tidal bridge and the tidal tail joint the disk of ngc6845a . since the age of the regions is 3 - 8myr
, star formation has been occurring along the tidal bridge and in the tidal arm well after they had begun to be torn apart ( @xmath0100myr ) .
satoh model fitting to the optical rotation curve of the a component reveals a kinematical mass of 4.4@xmath1 , inside the central 12kpc ( h@xmath2 ) .
the emission consist of two components , a more massive one that belongs to ngc6845a , and a second one associated to ngc6845b .
we do not detect gas associated to galaxies c and d. the total amount of is 1.4@xmath3 , five time higher than the content of the milky way .
the hi kinematics indicates an amount of dark matter associated to the a component two times higher than the mass inside its central 12kpc .
the group kinematics indicates an @xmath4 or @xmath5 , according to two different prescriptions for the internal absorption correction . in spite of this large difference ,
both values furnish similar group mass ( @xmath6 ) .
although preliminary results on n - body simulations indicate that either b or c might well create a tidal feature like the bridge of the a component , the collision with b appears to be more favourable . |
the critical phenomena in @xmath0he near the superfluid transition have been studied with exceedingly high precision of the temperature @xmath8 ( @xmath9 deg ) @xcite .
such experiments have thus been used to confirm or predict universal relations near the critical point . moreover , in this system , a number of unique nonequilibrium situations can be found in the nonlinear response regime . in particular , the superfluid transition is very sensitive to applied heat flow @xmath3 @xcite .
for example , if @xmath10 is above the transition temperature @xmath11 at one end of the cell and below @xmath12 at the other end , a hei - heii interface emerges separating superfluid and normal fluid regions @xcite . here
@xmath8 is nearly constant in the superfluid and has a finite gradient in the normal fluid , so the temperature gradient is almost discontinuous across the interface . in experiments , it emerges when @xmath0he in a normal fluid is cooled from a boundary below @xmath12 or when @xmath0he in a superfluid is warmed from a boundary above @xmath12 .
this interface has a thickness proportional to @xmath13 at small @xmath3 and is thus a unique singular object .
some authors were also interesred in the heat capacity in heat flow near the superfluid transition @xcite .
another interesting case is to quench liquid @xmath0he from its normal to its superfluid phase by a mechanical expansion @xcite . in such phase ordering with a multi - component order parameter , topological singularities ( vortices in @xmath0he ) should be proliferated in an early stage after quenching @xcite .
the superfluid transition can also be greatly influenced by gravity due to the pressure dependence of @xmath1 . in @xmath0he near the transition , the specific heat exhibits a sizable logarithmic singularity , but the isothermal compressibility has almost no singularity @xcite , resulting in nearly homogeneous density profiles for not large cell length on earth .
thus the height - dependence of @xmath1 is the main origin of inhomogeneity in equilibrium @xmath0he , preventing precise measurements of the critical phenomena in @xmath0he @xcite .
if the @xmath14 axis is taken in the upward direction on earth , @xmath1 decreases with the height @xmath14 as t_(p)= t _ ( 1 + gz ) , where @xmath15 is the transition temperature at the cell bottom and @xmath16@xmath17 at saturated vapor pressure with @xmath18 being the gravity constant .
the origin of the @xmath14 axis is at the cell bottom .
we introduce the local reduced temperature as /t_(p)-1 = ( t / t_-1 ) - g z , which depends on @xmath14 and represents the distance from the lambda line in the @xmath19-@xmath8 phase diagram therefore , equilibrium states are noticeably inhomogeneous in the temperature region || gh , where @xmath20 is the vertical cell length .
remarkably , in the temperature range @xmath21 at homogeneous @xmath8 ( in equilibrium ) , a gravity - induced two - phase coexistence is possible with an upper region in a superfluid and a lower region in a normal fluid @xcite .
the thickness @xmath22 and the characteristic reduced temperature @xmath23 of the gravity - induced interface is determined by the balance @xmath24 , where @xmath25 is the correlation length with @xmath26 .
hence they are given by _
g= _ + 0(_+0g)^-2/5 , _
g = ( _ + 0g)^3/5 , which are about @xmath27 and @xmath28 cm on earth , respectively .
it is worth noting that application of an electric field @xmath29 or a magnetic field @xmath30 gives rise to a shift of @xmath12 by an amount proportional to @xmath31 or @xmath32 .
if such a field is inhomogeneous , it leads to a gradient of @xmath12 , as well as the earth gravity @xcite .
furthermore , both gravity @xmath18 and heat flux @xmath3 can be crucial near the lambda line .
hereafter @xmath3 will be measured in units of erg@xmath6@xmath7s .
if heated from below , they both serve to decrease @xmath33 with increasing the height and the interface changes over from the gravity - induced one to the heat - induced one at @xmath34 with increasing @xmath3 .
( we may determine this crossover @xmath3 by equating the thicknesses of these interfaces . ) on the other hand , if heated from above , they can compete to produce new nonequilibrium states .
in fact , moeur _
et al._@xcite realized self - organized superfluid in @xmath0he heated from above , where the temperature gradient and the gradient of @xmath35 were balanced as t_= - t_g=0 , yielding a homogeneous @xmath36 with _
sos(q ) 1.010 ^ -8q^0.813 in the range @xmath37 .
this self - organized region extended over a macroscopic region in a 7.4 mm tall , 2 cm diameter cylindrical cell .
it is surprising that the distance @xmath33 to the lambda line was held fixed at an extremely small negative constant over a macroscopic region .
et al._@xcite pointed out that @xmath0he in heat flow is an example of self - organized criticality .
however , @xmath0he in heat flow and gravity stays slightly away from the lambda line , so it is not in a critical state , to be precise .
notice that the balance relation ( 5 ) itself follows from the vortex resistance in the gorter - mellink form @xcite , /dz
= t b(t ) q^3 , where @xmath38 with @xmath39 and @xmath40 near the transition @xcite .
the balance @xmath41 then yields @xmath42 with _
sos^v = ( b_0 q^3/g)^1/ 1.310 ^ -10q^1.35 . however , this @xmath43 is smaller than the experimental @xmath44 in eq.(6 ) by one or two orders of magnitude for @xmath45 . in previous simulations in one dimension @xcite , another kind of singular objects , phase slips , came into play , producing a thermal resistance in accord with the experiment . in the literature ,
phase slips were theoretically studied in one dimension @xcite and have been observed in quasi - one - dimensional superconductors @xcite and in @xmath0he in narrow apertures @xcite .
the self - organized superfluid states are analogous to the resistive states in wire superconductors in electric field ; however , in our case , phase slips appear as sheets even for macroscopic cell width .
this is because the uniaxial effect of gravity is intensified near the lambda line .
similar self - organization should well be realized in @xmath0he in the presence of a gradient of electric or magnetic field , for example , around the tip of a charged needle kept cooler than the surrounding liquid @xmath0he .
originally , the balance relation ( 5 ) was predicted for normal liquid @xmath0he @xcite .
liquid @xmath0he exhibits strong critical fluctuations of the complex order parameter slightly above the lambda line , which give rise to strong critical enhancement of the thermal conductivity @xmath46 @xcite . for small positive @xmath33
, experimental data of @xmath46 may be fitted to the following power law @xcite , _ 0 ^-x _ ( x_0.45 ) , where @xmath47 is a constant . in terms of @xmath33 in eq.(2 ) , the heat conduction equation reads c_p = , where the specific heat @xmath48 is treated to be a constant for simplicity . in steady states ,
the temperature gradient is @xmath49 so that /dz= ( q / t__0 ) ^x _ -g if @xmath50 , the balance in eq.(4 ) gives @xmath51 with _
son(q ) = ( t__0 g / q)^1/x _ = 3.810 ^ -9q^-2.22 .
this holds in the region @xmath52 , where @xmath53 is the relaxation length .
weichman and miller @xcite argued that the balance relation ( 5 ) is attained only when the correlation length @xmath25 is shorter than @xmath54 . from eq.(4 ) this local equilibrium condition is rewritten as @xmath55 .
thus we obtain the upper bound @xmath56 needed for the self - organization above the lambda line . for larger @xmath3 ,
the system crosses across the lambda line and a hei - heii interface appears . in accord with these results ,
et al._@xcite observed self - organized normal fluid states for @xmath57 . analyzing the nonlinear diffusive equation ( 10 ) ,
weichman and miller @xcite also predicted a new thermal wave propagating only upward in self - organized normal states .
this wave was observed both above and below the lambda line@xcite .
the physical processes in the self - organized superfluid remain quite unclear , since in the experiments near the transition measured was only the temperature and the simulations were one - dimensional .
realistic three - dimensional simulations are needed to investigate how vortices and phase slips come into play . with this purpose
, we will present dynamic equations in sec.ii and numerical results in three dimensions in sec.iii .
we use the renormalized model f equations @xcite accounting for gravity @xcite .
the dynamic variables are the complex order parameter @xmath58 and the entropy @xmath59 .
the superfluid density @xmath60 and the superfluid momentum density @xmath61 are proportional to @xmath62 and @xmath63 , respectively , where @xmath64 denotes taking the imaginary part .
the fourier components of @xmath65 and @xmath66 have wave numbers smaller than a upper cut - off @xmath67 .
the fluctuations with wave numbers larger than @xmath67 have already been coarse - grained@xcite . as @xmath67
is decreased down to the inverse correlation length @xmath68 , the renormalization effect is suppressed to justify use of the mean - field theory at long wavelengths , where the coefficients of the model are renormalized ones dependent on @xmath33 . in equilibrium superfluid
, the average order parameter behaves as @xmath69 . without gravity
, we define the correlation length as @xmath70 using @xmath26 .
we suppose a reference equilibrium superfluid state , where @xmath71 , @xmath72 , and @xmath73 with = 2.510 ^ -8 , = 1.610 ^ -3 cm .
hereafter we set @xmath74 and treat the fluctuations with sizes longer than @xmath75 .
those with sizes shorter than @xmath75 are little affected by heat flow .
the scaled position @xmath76 , the scaled time @xmath77 , the scaled correlation length @xmath78 , and the scaled reduced temperature @xmath79 will be written as @xmath80 , @xmath81 , @xmath82 , and @xmath83 in the same notation for simplicity , where 0.510 ^ -3 s is the thermal relaxation time on the scale of @xmath84 .
we introduce the scaled reduced temperatures , & & a(r , t)= ( t / t_bot-1)/ , + & & ( r , t ) = /= a- gz , where @xmath85 on earth . the equations for the scaled variables @xmath86 and @xmath87 read @xcite & & = i- l^ * , + & & m= a im(^*^2 ) + a , where @xmath88 is the scaled entropy deviation defined by m= a - a^2 ^ -1/2||^2/2 and @xmath89 is a universal number of order unity . the second term ( @xmath90 represents a decrease of the entropy in the ordered phase .
we set @xmath91 in our simulation . in gravity
the correlation length does not exceeds @xmath92 in eq.(4 ) , so we define the local correlation length as = _ g ( 1/_g ||^2/3 ) , where @xmath93 .
then we obtain two cases of @xmath94 and @xmath95 .
the scaled kinetic coefficients are written as @xcite l^*=b_^0.325 , ^ * = b_^0.675 .
we set @xmath96 and @xmath97 in our simulation @xcite . for @xmath0he near the lambda line ,
the complete hydrodynamic equations including the dynamic equation for @xmath65 were presented by the russian group many years ago @xcite . in the model
f @xcite , we do not treat the hydrodynamic equation for the mass density @xmath98 and the total momentum density @xmath99 neglecting the fluctuations of @xmath98 and @xmath100 , where @xmath101 is the normal fluid velocity and @xmath102 .
then the normal fluid velocity @xmath101 may be written as _
n = - ^-1 j_s .
therefore @xmath101 is much smaller than @xmath103 by the small factor @xmath104 .
this is justified for slow dynamic processes without first sounds near the lambda line . in critical dynamics , the dynamic equations have been treated as langevin equations with random source terms added @xcite . in this work , we neglect the random source terms in eqs.(17 ) and ( 18 ) , which produce the thermal fluctuations of the order parameter @xmath65 and the entropy @xmath66 .
they may be neglected when the small - scale thermal fluctuations shorter than @xmath82 have been coarse - grained in the dynamic equations .
the renormalization effect has already be accounted for in the renormalized coefficients . at @xmath105 after application of heat flux from the top at @xmath106 and @xmath107 @xmath108 ) .
the bulk region is gradually heated by repeated traversals of second sounds .
bottom : @xmath109 ( left ) and @xmath110 ( right ) at @xmath111 and @xmath112 .
a phase slip is being created at @xmath113 in the latter two times ( see the plates of video 1 ) . ] after application of heat flux from the top in a superfluid state in the same run as in fig.1 .
their colors represent the temperature deviation @xmath114 with the color map below the plates .
top : at @xmath115 a layer in the range @xmath116 is much disturbed , where a phase slip is going to appear .
temperature inhomogeneity is also apparent , for which see fig.3 .
middle : at @xmath117 a sheetlike phase slip is created near the bottom .
vortices above the sheet are expanding .
bottom : at @xmath118 the phase slip sheet is away from the bottom and is being broken into an aggregate of vortices .
video gives time evolution in the range @xmath119 . ]
in previous papers @xcite , eqs.(17 ) and ( 18 ) were integrated in one dimension , where phase slips play a major role in self - organized superfluid states , as in wire superconductors in electric field @xcite . in this work , we present numerical results in three dimensions on a @xmath120 cubic lattice with the mesh size equal to @xmath75 ,
so the corresponding cell length is @xmath121 mm .
the time increment @xmath122 of integration is @xmath123 in units of @xmath124s .
the periodic boundary conditions are imposed in the @xmath125 plane . at @xmath126 and @xmath127
, we set @xmath128 and assume homogeneity of the temperature and the heat flux in the @xmath125 plane , supposing metallic plates .
we prepared an equilibrium superfluid state with @xmath129 for @xmath130 .
we then applied a constant heat flux at @xmath131 , but we kept the bottom temperature at @xmath129 ( @xmath132 ) at @xmath126 for @xmath133 .
the scaled heat flux @xmath134 at the top is equal to 1 .
this corresponds to a heat flux of @xmath5 , which is the value of @xmath135 at @xmath136 . in the following
we will explain our numerical results . in fig.1 , we display profiles of the scaled reduced temperature @xmath114 vs @xmath14 ( with @xmath105 ) at seven times @xmath137 , and @xmath138 @xmath108 ) , where traversals of second sounds cause gradual heating of the bulk superfluid @xcite . since the bottom temperature is fixed , a transition layer supporting a temperature gradient grows near the bottom , where a heat flux of order 0.1 is from the liquid to the bottom plate .
this heat output is @xmath139 of the heat input from the top . in the lower plates , we show @xmath140 and @xmath141 near the bottom at three times to illustrate formation of a phase slip , where @xmath142 is the phase of @xmath143 .
we confirmed that this phase slip formation in a narrow region occurs even if the mesh size is reduced to @xmath144 . here
@xmath141 is equal to the superfluid velocity in the @xmath14 axis in units of @xmath145 , where @xmath146 is the @xmath0he mass .
note that the complex order parameter behaves as @xmath147 with @xmath148 in current - carrying superfluids and the bulk region is unstable for @xmath149 larger than @xmath150 from linear stability analysis@xcite . in our case ,
@xmath60 decreases towards zero and @xmath151 exceeds the critical value near the bottom , resulting in defect formation . in video 1
, we display the surfaces of @xmath152 near the bottom , which illustrates growth of vortices and phase slips . at @xmath115 disturbances of @xmath153
are enhanced in a layer in the range @xmath116 .
see the lower plates in fig.1 .
we can also see considerable inhomogeneity of the temperature . at @xmath117 ,
vortices are expanding and a sheetlike phase slip is at its birth . in our case under gravity , @xmath140 decreases near the bottom .
we observe that undulations on the sheet grow and detach to form vortices . at @xmath118 ,
the sheet is ramified into a vortex aggregate . afterwards , sheetlike phase slips successively appear forming an expanding self - organized superfluid . as well as vortex proliferation , vortex aggregates sometimes fuse into sheetlike phase slips .
these temporal changeovers occur on the system scale with large - amplitude second sounds .
no clear boundary can be seen between the self - organized and ordinary superfluid regions , but high - density vortices are undergoing large - scale temporal fluctuations in the transition region . in the @xmath154-@xmath155 plane at @xmath156 taken at @xmath115 in the same run as in fig.1 and video 1 , corresponding to the top plate in video 1 .
a phase slip is being created with considerable temperature inhomogeneity , where @xmath157 and @xmath158 with @xmath159 denoting the areal average . ] at @xmath160 , @xmath161 , and @xmath162 with @xmath105 after application of heat flux from the top in the time range @xmath163 in the same run as in figs.1 and 2 .
inset : @xmath140 vs @xmath14 at @xmath164 , which is zero at defect points . at this time ,
self - organized and ordinary superfluid regions are separated at @xmath165 . ] in fig.3 , the inhomogeneity of @xmath87 is shown in the @xmath154-@xmath155 plane at @xmath156 taken at @xmath115 .
it originates from the relation ( 16 ) for the entropy and the temperature .
that is , if @xmath60 increases in some regions , the temperature there tends to increase if the entropy change is slower . in fig.4 , time evolution of @xmath114
is shown at three heights with @xmath105 in a longer time range @xmath166 in the same run producing figs.1 and 2 .
for @xmath167 the three points are in an ordinary superfluid and are nearly uniformly heated .
subsequent zigzag temperature changes are due to the passage of platelike defect aggregates .
a similar plot was given by moeur _
@xcite , where the temperatures at three thermometers attached to the side wall became flat during the passage of a self - organized region . in our simulation ,
a train of defects eventually arrived at the top and a normal fluid with a steep temperature gradient emerged from the top . viewed from two angles in a dynamical steady state at @xmath168 .
their colors represent the temperature deviation @xmath114 with the color map below the plates .
a sheetlike phase slip can be seen at the middle , while the other defects are aggregates of vortices . ]
vs @xmath14 ( @xmath105 ) at @xmath169 , and 3910 in a dynamical steady state in the same run as in fig.4 .
heat fluxes at the top and bottom are fixed at a common value .
significant defect movements can be seen .
line of @xmath170 is also shown .
inset : @xmath114 vs @xmath14 at @xmath171 , 3840 , and 3850 in the ordinary superfluid region , indicating standing second sound waves . ] of @xmath172 ( upper plate ) in the range @xmath173 according to the color bar and @xmath174 ( lower plate ) in the @xmath154-@xmath14 plane at @xmath175 taken at @xmath176 , where use is made of the same data as in fig.5 . ] in fig.4 , the advancing speed of the self - organized region is about @xmath177 ( in units of @xmath178 or @xmath179 cm@xmath6s , while it was @xmath181cm@xmath6s in one of the experimental data ( see fig.2 of ref.19 ) .
note that the speed of this phase change is determined by the rate of heat input to the system . to check this aspect , we fixed the bottom heat flux at the top value 1 and resumed the integration using the data at @xmath182 in the run producing figs.1 - 4 .
then we could realize a dynamical steady state with a self - organized superfluid in a lower part and an ordinary superfluid in an upper part .
hereafter the times are 2500 plus those after the restart . in fig.5 , we display snapshots of the surfaces of @xmath183 at @xmath168 from above and side .
the defect structure here is composed of a single sheetlike phase slip in the middle and platelike vortex aggregates .
the vortices are highly connected one another forming a network .
some vortices start from the phase slip sheet or from the bottom wall . in this configuration , defects are absent in a layer region above the phase slip plane , but above this layer a vortex aggregate appears to form an interface between the self - organized and ordinary superfluid regions . in fig.6 , we present the scaled temperature deviation @xmath87 vs @xmath14 with @xmath105 at three consecutive times in the dynamical steady state .
it shows considerable displacive motions of the defect structure on a time scale of 70 .
the temperature variations are stepwise at a phase slip and are more gradual across vortex aggregates . in accord with the balance relation ( 5 ) ,
the zig - zag curves of @xmath114 are separated from the line of @xmath170 by 2.5 , which is our numerical result of @xmath184 . in agreement with this result , the experimental formula ( 6 ) yields _ sos/=11.2 ^ 0.813/2.5=2.85 .
note that phase slips in the one - dimensional simulation @xcite already gave @xmath44 in agreement with the experiment@xcite .
in the inset , we demonstrate the presence of second sounds oscillating on a time scale of 10 ( smaller than those in fig.1 ) in the upper ordinary superfluid region . in fig.7
, we show the phase @xmath185 and the gradient amplitude @xmath186 in the @xmath154-@xmath14 plane at @xmath175 using the same data as in fig.5 at @xmath176 .
these are cross - sectional profiles illustrating the complex defect structure .
around the defects , @xmath142 varies steeply in the upper plate , while peaks and ridges appear in the lower plate .
here we have @xmath187 summary and remarks -------------------- in summary , we have investigated the self - organized superfluid state on earth under a fixed downward heat flux @xmath5 erg@xmath6@xmath7s from the top in a cubic cell of @xmath121 mm length .
the periodic boundary condition has been imposed in the horizontal directions . first , at a fixed bottom temperature ,
we have examined how vortices and phase slips appear near the bottom and how a self - organized region expands upward .
second , by setting the bottom heat flux equal to the top heat flux , we have realized a dynamical steady state in which the self - organized and ordinary superfluids coexist .
the space - time fluctuations are highly enhanced , where a sheetlike phase slip often changes into a vortex aggregate and vice verse .
in fact , in fig.5 , we can see only a single sheetlike phase slip , while numerous vortex lines form platelike networks .
our numerical value of @xmath33 in eq.(2 ) ( the distance from the lambda line ) in self - organized superfluid states is in good agreement with the experimental @xmath44 in eq.(6 ) .
we estimate the temperature drop at a phase slip as ( t)_p= q_p/(_sos ) , where @xmath188 is the thickness of a phase slip ( @xmath189 ) and @xmath190 is the thermal conductivity in eq.(9 ) at @xmath191 .
if the defects consist of phase slips only , their density @xmath192 per unit length satisfies @xmath193 in self - organized superfluid states .
use of eqs.(9 ) , ( 12 ) , and ( 24 ) yields n_p_p = gt_(_sos ) /q
= ( _ son/_sos)^x_. this estimation is consistent with the results in the one - dimensional simulations @xcite . here
the required inequality @xmath194 holds for @xmath195 . in accord with this ,
self - organized superfluid states were observed only for @xmath195 in the experiment @xcite .
notice that the vortex aggregates in fig.5 also give rise to temperature drops comparable to that of a phase slip as in fig.6 .
a self - organized superfluid expands upward if the heat input from the top is larger than the heat output from the bottom .
after its arrival at the top , a normal fluid begins to expand downward ( not shown in this paper ) .
therefore , there can also be coexistence of a self - organized superfluid and a normal fluid .
this coexistence was realized in our previous simulation @xcite by cooling a normal fluid at the bottom below the transition . in such cooling
, a self - organized superfluid region grows upward to reach the top and then an ordinary superfluid region appears to expand downward . on heating or cooling from a boundary wall without gravity or at large @xmath3 , a hei - heii interface appears and its movement leads to the phase transformation between normal fluid and superfluid .
phase slips are well - known singular entities observed in quasi - one - dimensional superconductors @xcite and @xmath0he @xcite .
however , under an external potential , they can emerge as sheets in three - dimensional geometries .
the role of such a potential becomes relevant on approaching the criticality . in the case of gravity
, this crossover occurs extremely close to the lambda line .
self - organized superfluidity may be realized more easily in an inhomogeneous , intense electric or magnetic field , as stated in sec.i , or under high - speed rotation . in the literature@xcite ,
vortex generation after quenching of @xmath0he was proposed but it has not been detected in subsequent experiments , where crossing of the lambda line by a mechanical expansion is supposed . a similar quenching method ( using the piston effect )
was used to induce phase separation in fluids near the gas - liquid critical point @xcite . in a decompression experiment near the superfluid transition , however , we should examine how the phase transition is induced by sounds .
as recently reported @xcite , first sounds emitted from a boundary wall propagate throughout the cell with a stepwise pressure decrease at the wave front .
they are multiply reflected at the boundary walls until dissipated .
if the sound speed is @xmath196cm@xmath6s and the cell length is 1 cm , the acoustic traversal time is @xmath197s and is much longer than the relaxation time of the order parameter except extremely close to the lambda line .
second sounds are also emitted from the boundary walls with appearance of superfluid .
further investigation of these processes is thus needed .
one of the authors ( a.o . )
thanks r. duncan for stimulating discussions in an early stage of this work .
this work was supported by grants - in - aid for the global coe program
the next generation of physics , spun from universality and emergence " of kyoto university from the ministry of education , culture , sports , science and technology of japan .
r. nishida , j. taniguchi , r. nomura , and y. okuda , j. low temp .
phys . * 148 * , 109 ( 2007 ) . here
a hei - heii interface was visualized for @xmath198 .
a. onuki , j. low temp . phys . *
50 * , 433 ( 1983 ) ; ibid . *
55 * , 309 ( 1984 ) ; a. onuki and y. yamazaki , ibid .
* 103 * , 131 ( 1996 ) .
m. e. dodd , p. c. hendry , n. s. lawson , p. v. e. mcclintock , and c. d. h. williams , phys .
lett . * 81 * , 3703 ( 1998 ) ; v. b. efimov , o. j. griffiths , p. c. hendry , g. v. kolmakov , p. v. e. mcclintock , and l. skrbek , phys .
e * 74 * , 056305 ( 2006 ) .
w. j. skocpol , m. r. beasley , and m. tinkham , j. low temp . phys . * 16 * , 145 ( 1974 ) ; w. j. skocpol and m. tinkham , rep .
prog . phys .
1975 * 38 * , 1049 ( 1975 ) ; m. tian , j. wang , j. s. kurtz , y. liu , m. h. w. chan , t. s. mayer , and t. e. mallouk , phys .
b * 71 * , 104521 ( 2005 ) . | we investigate the nonlinear dynamics of @xmath0he slightly below the superfluid transition by integrating model f equations in three dimensions . when a superfluid is heated from above under gravity , a vortex tangle and sheetlike phase slips both appear near the bottom plate .
then a self - organized superfluid containing high - density vortices and phase slips grows upward , where high - amplitude second sounds are emitted from the self - organized to the ordinary superfluid region .
a phase slip sheet often changes into a vortex aggregate and vice versa .
the thermal resistance due to these defects produces a constant temperature gradient equal to the gradient of the pressure - dependent transition temperature @xmath1 . in this self - organized region ,
the temperature deviation @xmath2 consists of a negative constant determined by the heat flux @xmath3 and time - dependent fluctuations .
its time - average is calculated to be @xmath4 nk for @xmath5 erg@xmath6@xmath7s in good agreement with the experiment ( w.a .
moeur _
et al . _ ,
phys .
rev . lett . * 78 * , 2421 ( 1997 ) ) . |
much of our modern understanding of electronic correlations in narrow - band systems has derived from many - body treatments of model - hamiltonian systems such as the hubbard model and the anderson model .
for example , with respect to plutonium in particular , dynamical mean - field theory ( dmft ) approaches have been very useful in elucidating the physics of the very strong correlations in this material ( see , for example , refs .
@xcite and @xcite and references therein ) .
nonetheless , these calculations are not first principles , and most of the physics comes from the model part of the hamiltonian rather than the band - structure part of the calculations .
thus , it is still important to better understand the multi - orbital and hybridization effects in more realistic electronic - structure approaches that are less model dependent .
the gw method is our best modern tool to examine these effects , because it includes correlation effects beyond that of conventional local - density approximation ( lda ) band - structure techniques and yet is still first - principles . in this paper
we study correlation effects of fcc pu as a function of volume .
our goal is not to specifically elucidate the correlation physics of pu itself , since the gw approach is a low - order approximation and can not treat the very strong correlations of the @xmath1 phase of pu . rather , we wish to use the volume dependence to tune the material from high atomic density ( high pressure ) , where correlations can be greatly reduced due to the large hybridization between the pu @xmath3 orbitals , to low atomic densities ( where the pressure would actually be in tension ) , where the correlations effects are very strong . by following this procedure , we can understand how correlations modify the properties of a material @xmath4 . when the correlations become strong ,
it is certainly the case that higher - order approximations like dmft are necessary to accurately describe the material .
nonetheless , since gw is probably the correct starting point for such types of more sophisticated approaches , it is still useful to understand what happens to the electronic - structure of the material within the gw approximation as a function of the strength of the correlation effects , and , in particular , their effects on shifts in quasi - particle energies , which is what gw is best at representing .
in order to focus more specifically on the effects of correlations , we ignore one significant aspect of the electronic structure , viz . , the spin - orbit coupling , even though this is important for a detailed comparison with experiment .
spin - orbit coupling mainly shifts @xmath3 states around in energy , but at the same time add to the complexity of the individual @xmath3 bands , hence tending to hide some of the correlation physics in a bewildering array of bands .
spin - orbit effects can easily be added in when a more accurate comparison with experiment is desired ( as was done for an earlier paper on less correlated @xmath2 uranium @xcite ) .
for the same reasons we also ignore well - known large changes in crystal structure of pu metal with volume , and present calculations only for the simple fcc crystal structure as a function of volume .
we study a range of pu atomic volumes extending from well below that pertinent to the ground state @xmath2 phase to well above that of the high - temperature @xmath1 phase .
the actual crystal structure of @xmath2pu is a complicated monoclinic structure with 16 atoms per unit cell , while @xmath1pu has the fcc structure .
since @xmath1pu is well known to be a strongly correlated - electron metal while @xmath2pu appears to be reasonably well treated by conventional band - structure methods , we believe that our range of volumes corresponds to tuning the correlation effects between weak to moderate ( small volumes ) to strongly correlated ( large volumes ) . in the calculations to be presented in the following we mainly focus on changes in the effective bandwidth of the @xmath3 states in pu
. we will show that crystal - field effects are actually a more important factor in determining this bandwidth than the expected change in dispersion ( flattening of the bands ) .
the @xmath0 approximation can be viewed as the first term in the expansion of the non - local energy - dependent self - energy @xmath5 in the screened coulomb interaction @xmath6 . from a more physical point of view
it can be interpreted as a dynamically screened hartree - fock approximation plus a coulomb - hole contribution @xcite .
it is also a prescription for mapping the non - interacting green function onto the dressed green s function : @xmath7 .
this prescription can be described as follows . from the hamiltonian @xmath8 ( we use atomic rydberg units : @xmath9 , where @xmath10 and @xmath11 are the mass and charge of the electron ) @xmath12 may be constructed .
often @xmath13 is calculated from the lda eigenvalues and eigenfunctions ; however , there is no formal restriction for how to choose the initial starting point @xmath13 .
then , using the random phase approximation ( rpa ) @xcite , we can construct the polarization function @xmath14 and screened coulomb interaction @xmath6 as @xmath15 and @xmath16 where @xmath17 is the bare coulomb interaction .
the new green s function is defined as @xmath18 where @xmath19 is the potential due to ions ( madelung ) and @xmath20 is the hartree potential @xmath21 the single - particle density @xmath22 is calculated as @xmath23 .
in addition to this mapping of @xmath7 , one can also generate an excellent effective potential from @xmath24 that makes it possible to approximately do the inverse mapping of @xmath25 @xcite .
the is a method to specify this ( nearly ) optimal mapping of @xmath26 , so that @xmath27 can be iterated to self - consistency @xcite . at self - consistency
the quasiparticle energies of @xmath13 coincide with those of @xmath24 .
thus is a self - consistent perturbation theory , where the self - consistency condition is constructed to minimize the size of the perturbation .
the method is parameter - free , and independent of basis set as well as the lda starting point @xcite .
we have previously shown that reliably describes a wide range of @xmath28 @xcite , and rare - earth @xcite systems .
we have also applied the method to calculate the electronic structure of @xmath2-uranium @xcite .
our version of the method is based upon the full potential linear muffin tin orbital ( fp - lmto ) method @xcite , which makes no approximation for the shape of the crystal potential .
the smoothed lmto basis @xcite includes orbitals with @xmath29 ; both 7@xmath30 and 6@xmath30 as well as both 5@xmath3 and 6@xmath3 are included in the basis .
the @xmath31 orbitals are added as a local orbital @xcite , which is confined to the augmentation sphere and has no envelope function .
the @xmath32 orbital is added as a kind of extended ` local orbital , ' the ` head ' of which is evaluated at an energy far above fermi level @xcite and instead of making the orbital vanish at the augmentation radius a smooth hankel ` tail ' is attached to the orbital .
the 7@xmath30 and 6@xmath3 orbitals are necessary to obtain an accurate description of highlying bands , which are important for the accuracy of the polarization function in eq .
( [ polfct ] ) . for our calculations
we use the fcc pu lattice with the following lattice constants @xmath33=4.11 @xmath34 ( which corresponds to 90@xmath35 of the @xmath2-pu equilibrium volume ) , 4.26 @xmath34 ( at the @xmath2-pu equilibrium volume ) , 4.64 @xmath34 ( at the @xmath1-pu equilibrium volume ) and 4.79 @xmath34 ( at 110@xmath35 of the @xmath1-pu equilibrium volume ) .
-pu along two symmetry directions ( left panel ) , compared to ( b ) the lda energy bands ( right panel ) ; the mulliken weights of the @xmath3 orbitals are presented in red ( dark gray ) , for @xmath36 orbitals in black and for @xmath37 orbitals in blue ( light gray ) .
( c ) comparison of the total density of states ( dos ) for qsgw , red ( dark gray ) solid line , and lda , blue ( light gray ) dashed line .
the fermi energy is set at zero.,title="fig:",scaledwidth=45.0% ] -pu along two symmetry directions ( left panel ) , compared to ( b ) the lda energy bands ( right panel ) ; the mulliken weights of the @xmath3 orbitals are presented in red ( dark gray ) , for @xmath36 orbitals in black and for @xmath37 orbitals in blue ( light gray ) .
( c ) comparison of the total density of states ( dos ) for qsgw , red ( dark gray ) solid line , and lda , blue ( light gray ) dashed line .
the fermi energy is set at zero.,title="fig:",scaledwidth=45.0% ] in fig.1 we compare the qsgw one - particle electronic structure of @xmath1-pu with the lda band - structure results . the spin - orbit interaction is not included in this calculation .
the mulliken weights of the @xmath3 orbitals are presented in red ( dark gray ) , of the @xmath36 orbitals in black and of the @xmath37 orbitals in blue ( light gray ) . in both cases ,
the narrow bands located between -2 and 2 ev are predominantly due to seven 5@xmath3 orbitals . at the @xmath38 point
they are split by the cubic crystal field into one nondegenerate and two three - fold degenerate states .
this degeneracy is reduced at general k - points in the brillouin zone .
the lowest dispersive band centered around -4 ev has primarily @xmath36 character and the unoccupied bands above 2 ev are mainly due to @xmath37 orbitals . at the @xmath38
point the five @xmath37 states are split by the cubic crystal field into one two - fold degenerate and one three - fold degenerate state .
the @xmath39 hybridization along the symmetry directions presented in fig .
[ fig1 ] is generally very weak except for near the @xmath40 point along the @xmath41 direction , where there is very strong @xmath42 and @xmath43 hybridization for some of the @xmath37 and @xmath3 branches .
the degree of the hybridization is very similar in both the lda and @xmath44 calculations , as is the center of all of the bands .
the largest visible change is a significant narrowing of the @xmath45-band complex .
this effect has two components : first , the crystal - field splitting of the @xmath3 bands is significantly reduced in , second , the bandwidth of each individual @xmath3 branch ( band ) is also reduced in .
the total effect appears as an overall narrowing of the total density of states ( dos ) around the fermi level ( see fig .
[ fig1](c ) ) .
in addition , since the area under the curve is proportional to the number of @xmath45 states , which remains constant , the amplitude of the quasiparticle peaks are also higher in qsgw .
for example , the total dos at the fermi level in qsgw is 23 states / ev while in lda it is 9 states / ev .
orbitals , ( b ) the @xmath3 orbitals , and ( c ) the @xmath36 orbitals.,scaledwidth=55.0% ] the partial dos is presented in fig .
[ fig2 ] . in both calculations
the partial-@xmath45 dos is concentrated in a narrow energy interval around @xmath46 .
the partial-@xmath36 dos is mainly located in the occupied energy spectrum between -5 ev and @xmath46 and the @xmath37 bands are spread in a wide energy interval in both the unoccupied and occupied part of the spectrum with a few pronounced peaks at various energies .
overall , the and lda @xmath36 and @xmath37 peaks are located at the same energies , with the exception of the narrow peaks around @xmath46 , in which case the are visibly shifted closer to @xmath46 . in this region
these bands can be highly hybridized with the @xmath3 states , and thus these shifts reflect the band narrowing of the @xmath3 states .
the bottom of the @xmath36 and @xmath37 bands relative to the position of @xmath3 band is approximately at the same energy location in and lda .
the @xmath3 occupation changes in from its lda value of 5.06 to 4.85 states .
a similar reduction was observed in our calculations for @xmath2-uranium @xcite . in fig .
[ fig3 ] we present side by side the band structures for two different fcc volumes . on the left side is the band structure and partial @xmath36 and @xmath37 dos for the @xmath2 volume ( weakly correlated ) and on the right side is the band structure and partial @xmath36 and @xmath37 dos for the @xmath1 volume ( strongly correlated ) . in all cases ,
the red solid lines represent the results and the dashed - blue lines show the lda results .
two major effects are seen in this plot : ( 1 ) the @xmath3 bands narrow considerably with expanded volume , and ( 2 ) there is a similar effect on the @xmath37 bands ( notice the downward shift of the bands at the highest energy as one goes from the @xmath2 to the @xmath1 volume ) .
, @xmath47 are 1.02 , 1.37 ev in lda and 0.73 , 0.57 ev in qsgw .
@xmath47 is reduced significantly in qsgw calculation . for the 5@xmath3 state at the @xmath38 point , marked with a large dot , in fig .
[ fig4b ] we present the energy dependence of the self - energy and calculated spectral function @xmath48,scaledwidth=55.0% ] and @xmath49 ev.,scaledwidth=55.0% ] to examine band narrowing effects in more detail , in fig .
[ fig4 ] we expand the view of the 5@xmath3 bands .
also , in fig .
[ fig4b ] we show the self - energy and spectral function for one of the 5@xmath3 states . despite of the complicated energy dependence of the real and imaginary parts it appears that the state is described perfectly well by landau s fermi liquid theory .
the imaginary part of the self - energy goes through zero at @xmath46 and the spectral function has a _ single _ well defined peak centered at the eigenvalue .
there are no other pronounced features in the energy dependence of the spectral function .
this is representative of all @xmath45 states .
so the 5@xmath3 states within the theory are well defined quasiparticles with very large lifetime around @xmath46 ; the eigenvalues coincide with the quasiparticle energy .
therefore , our discussion is focused on the effects of electron correlations on the eigenvalues . at the @xmath38
point the @xmath3 orbitals are split by the cubic crystal field into a nondegenerate state @xmath50 and two three fold degenerate states @xmath51 and @xmath52 .
the splitting , @xmath53 , between the nondegenerate state and the lowest of the three - fold degenerate states is 1.02 ev in lda and 0.73 ev in .
the splitting , @xmath47 , between the two three - fold degenerate states is equal to 1.37 ev in lda but only 0.57 ev in .
this significant reduction of the crystal field splitting in is the major part of the band narrowing observed in the dos in fig .
another important aspect is the reduction of the width of each individual @xmath3 band . in table
[ table1 ] we present the values of the band width of pu @xmath45 bands along @xmath54 and @xmath41 symmetry directions .
the band width is defined as the difference between the maximum and the minimum energy of a particular band along the direction . in all cases
the band width is reduced significantly in .
the @xmath50 band along @xmath41 is a striking exception from this rule . as we show in fig .
[ fig5 ] this band strongly hybridizes with the @xmath37 states . at the @xmath38 point
it is 100@xmath35 @xmath3 but at the @xmath40 point it is predominately @xmath37 .
all the other bands shown in fig . [ fig4 ] remain 80 to 100@xmath35 @xmath3 throughout the symmetry directions shown in the figure .
this explains the anomalous change in the width of this band for this particular direction . from fig .
[ fig3 ] it is evident that , while in the width of the @xmath3 bands are significantly reduced , the width of @xmath37 bands are practically the same as for lda .
the @xmath50 band at the @xmath38 point has mainly @xmath3 character and therefore moves upward from its lda position due to the significant reduction of the @xmath3 crystal field in , but at the @xmath40 point mainly has @xmath37 character and therefore remains approximately at its lda position ( only slightly lower due to the slight downward shift of the center of @xmath37 band in ) .
the cumulative effect is that in this band is stretched .
-orbital weight of the @xmath50 band along @xmath41 direction .
also shown is the qsgw , red ( dark gray ) dashed line , and lda , blue ( light gray ) dashed line @xmath37-orbital weight of the @xmath50 band along the same direction , scaledwidth=55.0% ] [ table1 ] [ table2 ] [ table3 ] in tables [ table2 ] and [ table3 ] we show the values for the crystal field splitting of 5@xmath3 bands and @xmath3 band widths along @xmath54 for several volumes of the unit cell .
the band width and crystal field splitting of all bands is reduced as we move from the lower to higher volume case .
this is a result of the reduction of the @xmath3 band relative extend in the crystal .
it is also seen that the crystal field splittings and band widths for @xmath3 orbitals are always smaller than in lda .
we have also considered the band width of the entire 5@xmath3 band complex .
to do this we have applied a gaussian broadening on the partial-@xmath3 dos ( fig .
[ pdos ] ) . in this case
the 5@xmath3 band complex appears like a single large peak .
we can define the width of this band as the full width at half maximum ( fwhm ) of the peak .
this is also presented in table [ table3 ] .
it is evident that in both calculations the width is reduced as the volume increases .
the rate of reduction is the same in both calculations .
but the fwhm is always significantly smaller than the lda .
therefore , we conclude that in the @xmath3 orbitals contract due to the more accurate treatment of correlations . on the other hand ,
the @xmath36 and @xmath37 bands are very itinerant and are therefore already described accurately at the level of the lda treatment of correlations .
-partial dos for four different volumes .
the panels on the right side show the corresponding dos on the left panel broadened with a gauss function.,scaledwidth=65.0% ]
in conclusion , we have applied theory to @xmath1 plutonium as a function of volume in order to systematically understand the effects of electronic correlation on the band narrowing of the @xmath3 bands .
unlike conventional model - hamiltonian treatments of strongly correlated systems our approach is first principles and independent of any choice of model parameters , and hence provides a unique opportunity to examine the effect of electron correlations on the quasiparticle band structure as a function of @xmath3-orbital localization . in this way we have demonstrated that and lda prediction for the @xmath36 and @xmath37 electron subsystems are quite similar
this is because these electrons are very itinerant and therefore their description lies within the validity of lda .
however , the @xmath45 bands are much narrower than their lda counterpart .
we believe that our results for the first time show significant details of the band narrowing due to electron correlations that have not been previously studied . in particular , the calculations show that the major contribution to band narrowing is actually the reduction of the crystal - field splitting of 5@xmath3 states as compared to effects from a reduction in dispersion ( flattening of the individual bands ) .
this is a significant change in the character of the 5@xmath3 states and suggests the importance of using gw approaches as input to more sophisticated correlation approaches such as dynamical mean - field theories ( dmft ) .
this work was carried out under the auspices of the national nuclear security administration of the u.s .
department of energy at los alamos national laboratory under contract no .
de - ac52 - 06na25396 .
a.n.c would like to thank mark van schilfgaarde and takao kotani for informative discussions .
m. van schilfgaarde , t. kotani , and s. faleev , phys .
96 ( 2006 ) p. 226402
t. kotani , m. van schilfgaarde , and s. v. faleev , phys .
b 76 ( 2007 ) p. 165106
m. van schilfgaarde , t. kotani , and s. v. faleev , phys .
b 74 ( 2006 ) p. 245125
s. v. faleev , m. van schilfgaarde , and t. kotani , phys .
93 ( 2004 ) p. 126406
t. kotani , m. van schilfgaarde , s. v. faleev and a. n. chantis , j. phys . : condens .
matter ( 2007 ) p. 365236 a. n. chantis , m. cardona , n. e. christensen , d. l. smith , m. van schilfgaarde , t. kotani , a. svane , and r. c. albers , phys .
b 78 ( 2008 ) p. 075208
a. n. chantis , m. van schilfgaarde , and t. kotani , phys .
96 ( 2006 ) p. 086405
a. n. chantis , m. van schilfgaarde , and t. kotani , phys .
b 76 ( 2007 ) p. 165126
m. methfessel , m. van schilfgaarde and r. a. casali , in _ electronic structure and physical properties of solids : the uses of the lmto method _ , edited by h. dreysse , lecture notes in physics vol .
535 ( springer - verlag , berlin , 2000 ) | we present results for the electronic structure of plutonium by using a recently developed quasiparticle self - consistent @xmath0 method ( ) . we consider a paramagnetic solution without spin - orbit interaction as a function of volume for the face - centered cubic ( fcc ) unit cell .
we span unit - cell volumes ranging from 10% greater than the equilibrium volume of the @xmath1 phase to 90 % of the equivalent for the @xmath2 phase of pu .
the self - consistent @xmath0 quasiparticle energies are compared to those obtained within the local density approximation ( lda ) .
the goal of the calculations is to understand systematic trends in the effects of electronic correlations on the quasiparticle energy bands of pu as a function of the localization of the @xmath3 orbitals .
we show that correlation effects narrow the @xmath3 bands in two significantly different ways . besides the expected narrowing of individual @xmath3 bands ( flatter dispersion ) , we find that an even more significant effect on the @xmath3 bands is a decrease in the crystal - field splitting of the different bands . plutonium , electron correlations , gw approximation , first - principles electronic structure |
positronium ( ps ) , the bound state of an electron and a positron , is a purely leptonic system , and the triplet ( @xmath4 ) state of ps , orthopositronium(o - ps ) , decays slowly into three photons .
precise measurement of this decay rate gives us direct information about quantum electrodynamics(qed ) in bound state .
three precision measurements@xcite of the o - ps decay rate were performed at ann arbor , which reported decay rate values much larger , i.e. , 5.2 9.1 experimental standard deviations , than a qed prediction@xcite ( @xmath5 ) corrected up to @xmath3 .
this discrepancy has been referred as ` orthopositronium lifetime puzzle ' , and was long - standing problem . to elucidate discrepancies ,
a variety of experiments have since been carried out to search for the exotic decay mode of o - ps , resulting in no evidence so far @xcite . as some fraction of o - ps inevitably results in `
pick - off ' annihilations due to collisions with atomic electrons of the target material , the observed o - ps decay rate @xmath6 is a sum of the intrinsic o - ps decay rate @xmath7 and the pick - off annihilation rate into @xmath1 s , @xmath8 , i.e. , @xmath9 @xmath10 is proportional to the rate of o - ps collisions with the target materials , i.e. ; @xmath11 , where @xmath12 is product of the density of the target , @xmath13 the annihilation cross - section , and @xmath14 the time dependent velocity of o - ps . due to the thermalization process of o - ps , this necessitates expressing @xmath8 as a function of time whose properties are dependent on the surrounding materials .
thermalization process should be carefully treated even in the cavity experiment@xcite .
although pickoff correction is small in cavities , disappearance of o - ps through the cavity entrance aperture has large contribution to @xmath6 .
this disappearance rate is also proportional to @xmath14 , as the same reason .
since the rate of elastic collision is extremely small in cavities , it takes much time , longer than 1 @xmath15 , to thermalize well , and the disappearance rate still depends strongly on time . in previous measurements@xcite
, @xmath6 s were measured by varying the densities of the target materials , size of the cavities and also the entrance aperture of the cavities .
the extrapolation to zero density or aperture was expected to yield the decay rate in a vacuum , @xmath16 , under the assumption of quick thermalization ( shorter than 170 - 180 nsec ) with constant o - ps velocity .
however , this assumption contains a serious systematic error as pointed out in reference@xcite .
we have proposed the following entirely new method@xcite , which is free from above - mentioned systematic error .
the energy distribution of photons from the 3-body decay is continuous below the steep edge at 511 kev , whereas the pick - off annihilation is 2-body which produces a 511 kev monochromatic peak .
energy and timing information are simultaneously measured with high - energy resolution germanium detectors such that @xmath17 can be determined from the energy spectrum of the emitted photon .
once a precise thermalization function is obtained , @xmath10 will contain all information about the process .
the population of o - ps at time @xmath18 , @xmath19 can be expressed as @xmath20 providing the ratio is determined as a function of time , the intrinsic decay rate of o - ps , @xmath16 , can be directly obtained by fitting the observed time spectrum .
we obtained decay rate of @xmath21 and @xmath22 independently@xcite , which are consistent with the non - relativistic qed calculation@xcite , and quite differ from the results obtained at ann arbor@xcite , 7.0482(16)7.0516(13)@xmath23 .
the observed @xmath10 indicates that o - ps thermalization is slow and it is serious systematic problem in all experiments using an extrapolation . in 1998
, they recognized that the incomplete thermalization makes the serious problem in their results@xcite , but they did not update results , and the discrepancy still remaining .
there are also still several problems remained in our result@xcite : ( i ) accuracy was @xmath24 , being worse than those of the other experiments@xcite , ( ii ) there was unknown systematic uncertainties before @xmath25 , and therefore to remove this uncertainty , final results were obtained using data after @xmath26 , and ( iii ) systematic error regarding the stark effect was not estimated .
improving results by considering these problems is described@xcite here , and we report the final result using our new method .
figure [ fig : setup ] shows a diagram of the experimental setup ; a @xmath27 positron source ( dia .
, @xmath28 ) with approximate strength of @xmath29 , being sandwiched between two sheets of plastic scintillators and held by a cone made of aluminized mylar .
the cone was situated at the center of a cylindrical @xmath30-diameter vacuum container made of @xmath31-thick glass , being filled with @xmath32 powder and evacuated down to @xmath33 torr .
two different types of @xmath0 powder were used as listed in table [ table : powder ] , with the biggest difference being in the mean distance between grains such that different pick - off ratio would be obtained . using these powders ,
two 6-month runs were performed .
.characteristics of @xmath0 powders used in the measurements . [ cols="^,>,>",options="header " , ] table [ table : systematic ] summarizes estimates of various systematic errors , and the details are shown in reference@xcite .
we just mention here about errors related to simulation and stark effect .
monte carlo simulation related errors : the predominant contribution to total systematic error is produced by uncertain normalization .
that is , the number of pick - off events are determined by subtracting the normalized 3-@xmath34 spectrum of monte carlo simulation from the o - ps spectrum , where changing the normalization factor affects the @xmath17 values and eventually propagates to the final result . since the sharp fall - off of the 3@xmath34-spectrum at 511 kev
is solely produced by the good ge energy resolution of @xmath35 kev , this subtraction only affects the lower side of the pick - off spectrum such that improper subtraction results in asymmetry of the pick - off spectrum shape .
comparison of the asymmetries of the pick - off peak shape and the prompt peak annihilation spectrum is a good parameter for estimating this systematic error .
the 1 @xmath36 error is assessed as @xmath37 .
inhomogeneity of the powder affects on the detection efficiency .
@xmath0 powder density in the mc simulation is conservatively changed by @xmath38 although the uniformity is known to be within a few % ; a change resulting in an error of @xmath39 ppm .
the stark shift stretches the lifetime of ps atoms , i.e. , a perturbative calculation shows that the shift is proportional to a square of the effective electric field @xmath40 such that @xmath41 , where @xmath42 .
@xmath40 is defined as the root - mean - square electric field sensed by o - ps during its lifetime .
calculations have estimated two contributions exist based on measurements of the electrical charge - up on the primary grains of silica powders and electrical dipole moment on the surface of grains@xcite .
the charge - up is partly intrinsic depending on powder specifications and partly due to positron depositions from the @xmath43 source .
the effect , however , is negligible in both cases , i.e. , on the level of @xmath44 at most .
silanol functional groups on the surface of the powder grain behave as an electrical dipole moment creating an effective field around the grains .
average densities are measured to be @xmath45 and @xmath46 for run 1 and 2 , respectively .
accordingly , the effective field can be analytically calculated such that the contribution to the o - ps decay rate is determined to be @xmath47 for run1 and @xmath48 for run2 .
these estimations were confirmed using results from precise hyper - fine - structure ( hfs ) interval measurements of ground state ps in silica powder@xcite , where the interval is proportional to the size of stark effect . considering the difference in powder densities used , the hfs results are consistent with our estimation .
other sources of systematic errors : error contribution due to the zeeman effect is estimated using the measured absolute magnetic field around the positronium assembly ( @xmath49 ) .
since the 3-@xmath34 pick - off process can only occur at a certain ratio , the calculated relative frequency @xmath50 is consistent with previous measurements@xcite , being @xmath51 for run1 and @xmath52 for run2 .
the probability of the excited state ( n=2 ) of ps is about @xmath53 @xcite , which could make the intrinsic decay rate @xmath54 smaller due to the low decay rate ( @xmath55 ) the above discussed systematic errors are regarded as independent contributions such that the total systematic error can be calculated as their quadratic sum , resulting in @xmath56 , @xmath57 for run1 and @xmath58 , @xmath59 for run2 .
-corrected nrqed prediction .
small vertical lines on the error bars indicate the size of errors solely due to statistics while error bars represent total ambiguities including systematic errors . ]
the decay rate of o - ps formed in @xmath0 was measured using a direct @xmath1 correction method in which the thermalization effect of o - ps is accounted for and integrated into the time spectrum fitting procedure .
consistent results were obtained using two runs with different types of @xmath0 powders , i.e. , and a weighted average gave @xmath60 , which agrees well with our previous result@xcite , it disagrees with results obtained at ann arbor , i.e. , by 3.8@xmath61 @xcite . as illustrated in fig .
[ fig : history ] , our value agrees well with the nrqed prediction corrected up to @xmath3 term@xcite .
we confirm our results obtained at 1995 and 2000 , and improve accuracy by factor of 1.8 .
just one week before this workshop , ann arbor reported new result@xcite using nanoporous silica film , which produces near - thermal energy o - ps .
they recognized the systematic problem due to unthermalized o - ps ( fast o - ps ) and suppress the kinetic energy of the emitted o - ps using nanoporous silica film and positron beam energy .
they obtain value of @xmath62 , which completely agrees with our results as shown in fig.6 .
so we can conclude that the unthermalized o - ps has contributed large systematic errors of about 1000 ppm and made discrepancy between the qed prediction as pointed out in references@xcite .
now two groups ( two methods ) obtain the consistent results and the orthopositronium lifetime puzzle is clearly solved here .
the combined results of two different methods is @xmath63 , which is not enough accuracy(140ppm ) to examine the @xmath3-correction ( 210ppm ) .
we need much effort to improve accuracy .
we wish to thank prof .
s. orito , t. hyodo and y. nagashima ( univ . of tokyo ) for very useful suggestions for discussions .
it is a pleasure to thank prof .
adkins ( franklin & marshall college ) for calculating the @xmath64 corrected energy spectrum .
lett . _ * b351 * 574 ( 1995 ) ; + s. asai , k. shigekuni , t. sanuki , and s. orito , _ phys .
_ * b323 * 90 ( 1994 ) ; + m. tsuchiaki , s. orito , t. yoshida , and m. minowa , _ phys .
lett . _ * b236 * 81 ( 1990 ) 81 .
, it euro .
. lett . * 33 * 111 ( 1996 ) ; + a. badertscher , p. crivelli , m. felcini , s.n .
gninenko , n.a .
goloubev , p. nedelec , j.p .
peigneux , v.postoev , a. rubbia and d. sillou _ phys .
_ * b542 * 29 ( 2002 ) , `` study of bound state qed : precision measurement of the orthopositronium decay rate '' , ph .
d. thesis , university of tokyo ( 2001 ) ; + o. jinnouchi , s. asai , t. kobayashi `` precision measurement of orthopositronium decay rate using @xmath0 powder '' accepted by _ phys .
b_. , _ j. phys .
_ * b31 * 329 ( 1998 ) ; + y. nagashima , m. kakimoto , t. hyodo , k. fujiwara , a. ichimura , t. chang , j. deng , t. akahane , t. chiba , k. suzuki , b. t. a. mckee , and a. t. stewart , _ phys .
rev . _ * a 52 * 258 ( 1995 ) . | the intrinsic decay rate of orthopositronium formed in @xmath0 powder is measured using the direct @xmath1 correction method such that the time dependence of the pick - off annihilation rate is precisely determined .
the decay rate of orthopositronium is found to be @xmath2 , which is consistent with our previous measurements with about twice the accuracy .
results agree well with the @xmath3 qed prediction , and also with a result reported very recently using nanoporous film . `
ut - icepp 03 - 06 + ` july 2003 + * solution of orthopositronium lifetime puzzle * contribution to proceedings of + workshop on positronium physics + eth honggerberg zurich ( may 30 - 31 2003 ) + s. asai , o. jinnouchi and t. kobayashi icepp international center for elementary particle physics , university of tokyo 7 - 3 - 1 hongo , bunkyo - ku , tokyo 113 - 0033 , japan |
the argo - ybj experiment is under way over the next few years at yangbajing high altitude cosmic ray laboratory ( 4300 m a.s.l .
, 606 @xmath4 ) , 90 km north to lhasa ( tibet , p.r .
the aim of the experiment is the study of cosmic rays , mainly cosmic @xmath5-radiation , at an energy threshold of @xmath6 @xmath7 , by means of the detection of small size air showers at high altitude .
the apparatus consists of a full coverage detector of dimension @xmath8 realized with a single layer of resistive plate counters ( rpcs ) .
the area surrounding the central detector core , up to @xmath9 , consists of a guard ring partially ( @xmath10 ) instrumented with rpcs .
these outer detector improves the apparatus performance , enlarging the fiducial area , for the detection of showers with the core outside the full coverage carpet .
a lead converter @xmath11 @xmath12 thick will cover uniformly the rpc plane in order to increase the number of charged particles by conversion of shower photons and to reduce the time spread of the shower front . the site location ( longitude @xmath13 31 50 e , latitude @xmath14 06 38
n ) permits the monitoring of the northern hemisphere in the declination band @xmath15 .
such a detector , performing a continuous high sensitivity sky survey , complements the narrow field of view air cerenkov telescopes allowing to bridge the gev and tev energy regions and to face a wide range of fundamental issues in cosmic ray and astroparticle physics including @xmath5-ray astronomy , grbs physics and the measurement of the @xmath16 at tev energies @xcite .
detector assembling will start late in 2000 and data taking with the first @xmath0 750 @xmath17 of rpcs in 2001 . in order to investigate both the rpcs performance at 4300 @xmath18 a.s.l . and the capability of the detector to sample the shower front of atmospheric cascades , during 1998 a full coverage carpet of @xmath19 has been put in operation in the yangbajing ( ybj ) laboratory .
results concerning both the rpcs behaviour and the features of the showers imaged by the carpet are presented .
the set - up of the detector is an array of @xmath20 chambers of area @xmath21 @xmath22 each , covering a total area of @xmath23 @xmath17 .
the active area of @xmath24 @xmath17 , accounting for a dead area due to a @xmath25 @xmath26 frame closing the chamber edge , corresponds to a @xmath27 coverage .
the rpcs , with a @xmath28 @xmath26 gas gap , are built with bakelite electrode plates of volume resistivity in the range ( @xmath29 ) @xmath30 .
the rpc signals are picked up by means of aluminum strips @xmath31 @xmath12 wide and @xmath32 @xmath12 long which are glued on a @xmath33 @xmath26 thick film of poly - ethylen - tereftalate ( pet ) . at the edge of the detector
the strips are connected to the front - end electronics and terminated with 50 @xmath34 resistors .
a grounded aluminum foil is used to shield the bottom face of the rpc and an extra pet foil , on top of the aluminum , is used as a further high voltage insulator .
the front - end circuit contains 16 discriminators , with about @xmath35 @xmath36 voltage threshold , and provides a fast - or signal with the same input - to - output delay ( @xmath37 @xmath38 ) for all the channels .
this signal is used for time measurements and trigger purposes in the present test .
the 16 strips connected to the same front - end board are logically organized in a pad of @xmath39 @xmath40 area .
each rpc is therefore subdivided in 10 pads which work like independent functional units .
the pads are the basic elements ( `` pixel '' ) which define the space - time pattern of the shower ; they give indeed the position and the time of each detected hit . at any trigger occurrence
the times of all the pads are read - out by means of multihit tdcs of @xmath41 @xmath38 time bin , operated in common stop mode .
since the pad signal is shaped to 1.5 @xmath42 , only the time profile of the earliest particles hitting the pads is imaged .
the set - up was completed with a small telescope consisting of 3 rpcs of @xmath43 @xmath40 area with 16 pick - up strips @xmath44 @xmath12 wide connected to front - end electronics board similar to the ones used in the carpet .
the 3 rpcs were overlapped one on the other and the triple coincidence of their fast - or signals was used to define a cosmic ray crossing the telescope .
the rpcs were operated in streamer mode as foreseen for the final experiment .
this mode delivers large amplitude saturated signals and is less sensitive than the avalanche or proportional mode to electromagnetic noise , to changes in the environment conditions and to mechanical deformations of the detector .
three gas mixtures were tested which used the same components , argon , isobutane and tetrafluoroethane , in different proportions : tfe / ar / i - but = 45/45/10 ; 60/27/13 and 75/15/10 . an higher tfe concentration in place of the ar concentration increases the primary ionization thus compensating for the @xmath45 reduction caused by the lower gas target pressure ( 600 @xmath46 ) and reduces the afterpulse probability .
the reduction of the argon concentration in favour of tfe results in a clear increase of the operating voltage as expected from the large quenching action of tfe . since the higher the tfe fraction
, the lower is the charge delivered in the gas by a single streamer , we decided to operate the test carpet with the gas mixture corresponding to the highest tfe fraction , in order to extend the dynamic range achievable for the analogical read - out .
[ effic ] shows the operating efficiency for 4 ored pads .
the efficiency was measured using cosmic ray signals defined by means of an auxiliary telescope placed on top of the carpet .
the same curve for a @xmath28 @xmath26 gap rpc operated at sea level is also shown for comparison .
the detection efficiency @xmath47 operating voltage , compared to the operation at @xmath48 @xmath46 pressure in ybj , shows an increase of @xmath49 @xmath50 in operating voltage .
the effect of small changes in temperature @xmath51 and pressure @xmath52 on the operating voltage can be accounted for by rescaling the applied voltage @xmath53 according to the relationship : v = v_a where @xmath54 and @xmath55 are arbitrary standard values , e.g. @xmath56 @xmath46 and @xmath57 @xmath58 respectively for a sea level laboratory .
this formula predicts , starting from the ybj data , an operating voltage at sea level which is considerably smaller than the experimental one .
however , a good consistency is recovered by assuming that , in the ideal gas approximation , the parameter which fixes the operating voltage is given by _
gap @xmath59 pressure / temperature_. measurements on rpcs with different gas gap size justify this assumption .
[ effic ] also shows that the plateau efficiency measured at ybj is @xmath60 lower than at the sea level .
although a lower efficiency is expected from the smaller number of primary clusters at the ybj pressure , we attribute most of the difference to the underestimation of the ybj efficiency . at ybj altitude , indeed , the ratio of the cosmic radiation electromagnetic to muon component is @xmath61 times larger than at sea level .
a spatial tracking with redefinition of the carpet track downstream would eliminate the contamination from soft particles , giving a more accurate and higher efficiency . on the other hand the lower efficiency
could hardly be explained with the gas lower density .
in fact , the number of primary clusters in the ybj test , estimated around @xmath62 , is the same as in the case of some gas , e.g. ar / i - but / cf3br = 60/37/3 , that was frequently used at sea level with efficiency of @xmath63 . the counting rate of 4 ored pads , together with the rpc current ,
are reported in fig .
[ countrate ] @xmath47 the operating voltage .
the results for a percentage of @xmath64 of tfe are also reported for comparison . a rather flat singles counting rate plateau is observed at a level of @xmath65 @xmath66 for a single pad of area @xmath67 @xmath40 .
the time jitter distribution of the pad signals was obtained by measuring the delay of the fast - or signal with respect to rpc2 ( the rpc in the central position of the trigger telescope ) by means of a tdc with 1 @xmath38 time bin .
this distribution is shown in fig .
[ timej ] for four pads .
the average of the standard deviations is @xmath68 @xmath38 corresponding to a resolution of @xmath69 @xmath38 for the single rpc if we account for the fact that the distributions show the combined jitter of two detectors . a more detailed presentation of these results is reported in @xcite .
data were taken either with or without a 0.5 @xmath12 layer of lead on the whole carpet in order to investigate the converter effect on multiplicity and angular resolution .
a trigger based on pad multiplicity has been used to collect @xmath70 shower events in april - may 1998 . the integral rate as a function of the pad multiplicity is shown in fig .
[ freq ] for showers before and after the lead was installed .
a comparison at fixed rate indicates an increase of pad multiplicity due to the effect of the lead of @xmath71 , as expected according to our simulations . since in this test
the strip read - out is not effective , the actual multiplicity is understimed for particle densities higher than @xmath69 @xmath72 .
data have been corrected to obtain the average particle density as a function of the recorded multiplicity ( fig .
[ densit ] ) . due to the reduced size of the detector
, the time profile of the shower front sampled by the carpet is expected to exhibit a planar shape . in this approximation
the expected particle arrival time is a linear function of the position .
we performe an optimized reconstruction procedure of the shower direction as follows : * unweighted plane fit to hits for each event by minimization of the function ^2 = _ i\{lx_i + my_i + nz_i - c(t_i - t_0)}^2 the sum includes all counters with a time signal @xmath73 , @xmath74 is the light velocity , @xmath75 are the @xmath76th center pad position coordinates .
the parameters of the fit are the time offset @xmath77 and the @xmath78 direction cosines . *
rejection of outlying points by means of a @xmath79 cut and iteration of the fit until all points verify this condition . if the remaining hits number is @xmath80 the event is rejected . here
@xmath81 is the standard deviation of the distribution .
this procedure is rather fast because it makes use only of analytic formulae .
no _ a priori _ information about shower features is required .
the actual value of @xmath58 could depend on the features of the reconstructed showers as well as on the experimental conditions ( pad dimension , shaping of the signals , multihit capability , etc . ) . by choosing @xmath82 about @xmath83 of the hits
are rejected .
increasing @xmath58 should increase the number of discarded hits without a significant improvement of the @xmath84 .
the @xmath84 distribution exhibits a long tail more pronounced for low multiplicity events . in these high @xmath84 events
the time hits are considerably spread , by far more than expected from the detector time resolution .
they can be attributed to the sampling of a portion of the shower affected by large time fluctuations .
events with reduced @xmath85 @xmath86 ( @xmath87 of the total ) have been discarded and not used in the following analysis .
since the shower size is not measured , events have been classified according to the observed particle density .
the zenith angle distribution of the integral density spectrum is expected to follow an exponential behaviour i(,)=i(,0)e^ -x_0 ( sec-1 ) where @xmath88 is the vertical depth .
the attenuation mean free path @xmath89 of eas is related to the absorption length @xmath90 of eas particles by @xmath91 where @xmath5 is the index of the integral density spectrum .
the angular distribution of events with density greater than 3.0 particles/@xmath17 ( corresponding to @xmath92 = @xmath93 ) is shown in fig .
[ zenith ] .
the data can be fit out to @xmath94 with an @xmath95 $ ] law .
the parameter @xmath96 is found to be @xmath97 , so that @xmath98 @xmath4 , in excellent agreement with previous results concerning the density spectrum of shower particles @xcite . for angles greater than @xmath99 a deviation from this law
is observed .
misreconstructed events , showers locally produced in the walls of the sourrounding buildings and horizontal air showers could contribute to these large angle events .
this result is substantially unchanged for densities ranging from 1 to 5 particles/@xmath17 .
data confirm that the shape of the angular distribution is dominated by the physical effect of atmospheric absorption and that no relevant instrumental effects or selection biases have been introduced .
the differential density spectrum measured in three different intervals of zenith angles is shown in fig .
[ densthet ] .
the angular bin size is @xmath100 , so that the difference of atmospheric depth inside the source bin is @xmath101 @xmath4 ( about one radiation length ) for zenith angles up to @xmath102 .
a smaller bin size has been used to analyze data up to @xmath99 .
the spectra refer to the depth corresponding to the mean zenith angle of the showers recorded inside each interval .
the shapes of the spectra are very similar in the density range @xmath103 particles/@xmath17 not affected by threshold or saturation effects .
the index @xmath104 of the density spectrum as a function of the zenith angle is shown in fig . [ spectra ] . a weighted mean up to @xmath105 gives @xmath106 .
this value is fully consistent with the ones obtained in experiments at high altitude at comparable particle densities @xcite .
montecarlo simulations show that present data concern energies ranging from about 1 tev ( quasi - vertical showers ) to a few hundred tev ( @xmath107 ) .
accordingly , the measured @xmath104 can be considered a reasonable estimate of the slope of the density spectrum of showers initiated by primaries ( mainly protons ) of energies @xmath108 @xmath109 . from these data
the absorption length of eas particles @xmath90 is found to be @xmath110 @xmath4 consistent with results of similar measurements at small shower sizes @xcite .
data recorded at large zenith angles ( @xmath111 ) are due to showers developed through atmospheric depths greater than @xmath112 @xmath4 .
thus , a substantial contribution is expected from showers originated by primaries of energies beyond the knee of the cosmic ray spectrum .
this could explain the increasing of the slope @xmath104 with zenith angle , as shown in fig .
[ spectra ] .
the time distribution of the shower hits with respect to the fitted plane is shown in fig .
[ timethick ] for two different multiplicity ranges .
quasi - vertical showers ( @xmath113 ) have been selected .
the distribution of time residuals exhibits , as expected , a long tail due to time fluctuations and to the curved profile of the shower front , more pronounced for low multiplicity .
the width of these distributions is related to the time thickness of the shower front .
since the position of the shower core is not reconstructed , the experimental result concerns a time thickness averaged on different radial distances . for low multiplicity events with a mean number of particles per pad @xmath114 , the spread @xmath81 represents a reasonable measurement of the shower thickness . for high multiplicity events ( @xmath115 ) , where the mean number of particles hitting one pad is @xmath116
, this @xmath81 does not express the thickness of the shower disc without any bias .
in fact , the time residual distribution is related to the fluctuations of the first particle detected by each pad . taking into account the total detector resolution of @xmath117 @xmath38 ( rpc intrinsic jitter , strip length , electronics time resolution ) the time jitter of the earliest particles in high multiplicity events ( @xmath118 hits )
is estimated @xmath69 @xmath38 .
the averaged shower front thickness of low multiplicity events ( particle density @xmath119 @xmath72 ) is found @xmath120 @xmath38 .
the angular resolution of the carpet has been estimated by dividing the detector into two independent sub - arrays ( `` odd pads '' and `` even pads '' ) and comparing the two reconstructed shower directions .
these two sub - arrays overlap spatially so that they are affected by the same shower curvature .
events with total number of hits @xmath121 have been selected according to the constraint @xmath122 . the distribution of the even - odd angle difference @xmath123 is shown in fig .
[ evenodd ] for events in different multiplicity ranges and @xmath124 .
these distributions narrow , as expected , with increasing shower size .
the effect of the lead sheet on the angular resolution can be appreciated in fig .
[ median ] , where the median @xmath125 of the distribution of @xmath123 as a function of pad multiplicity , for showers reconstructed before and after the lead was added , is shown .
the improvement of the angular resolution is a factor @xmath0 1.4 for @xmath126 and decreases with increasing multiplicity . assuming that the angular resolution function for the entire array is gaussian , its standard deviation @xmath81 is given by @xcite @xmath127 this chessboard method yields a standard deviation @xmath128 for events with a pad multiplicity @xmath129 .
a carpet of @xmath0 50 @xmath17 of rpcs has been put in operation at the yangbajing laboratory in order to study the high altitude performance of rpcs and the detector capability of imaging with high granularity a small portion of the eas disc , in view of an enlarged use in tibet ( argo - ybj experiment ) .
the results of this test confirm that rpcs can be operated efficiently ( @xmath130 ) to sample air showers at high altitude with excellent time resolution ( @xmath69 @xmath38 ) .
the analysis of data collected with a shower trigger suggest that the rpcs carpet capability of reconstructing the shower features is consistent with expectations . as a conclusion , the overall results of the test look well promising for future operation of the full detector .
9 abbrescia m. et al . , _ astroparticle physics with argo _ , proposal ( 1996 ) .
+ download this document at the url : www1.na.infn.it/wsubnucl/cosm/argo/argo.html bacci c. et al . ,
submitted to nucl .
bennet s. et al . , j. phys .
soc . of japan ,
a - iii ( 1961 ) 196 .
greisen k. , progr .
cosmic ray physics , vol .
iii ( 1956 ) .
alexandreas d.e .
et al . , nucl .
instr . meth .
a311 ( 1992 ) 350 . | an rpc carpet covering @xmath0 10@xmath1 m@xmath2 ( argo - ybj experiment ) will be installed in the yangbajing laboratory ( tibet , p.r . china ) at an altitude of 4300 m a.s.l .. a test - module of @xmath0 50 m@xmath2 has been put in operation in this laboratory and about 10@xmath3 air shower events have been collected . the carpet capability of reconstructing the shower features is presented . |
the hera collider , which operated at desy from 1992 to 2007 , has produced many important physics results , first of all perhaps the most precise determinations to date of the quark and gluon densities in the proton from single experiments ( h1 , zeus ) and their combined data sets @xcite .
these data , taken in deep - inelastic electron - proton scattering , are complemented by a wealth of data from photoproduction at low virtuality of the exchanged photon , in particular on jet production , giving access also to the distributions of partons in the photon and to measurements of the strong coupling constant @xcite . using the full data set of the hera run from 2005 - 2007 with an integrated luminosity of 300 pb@xmath0 ,
the zeus collaboration have recently published a final measurement of inclusive jet photoproduction @xcite and used it to determine the strong coupling constant ( at the mass @xmath1 of the @xmath2-boson ) to be _
s(m_z)=0.1206^+0.0023_-0.0022 ( exp . ) ^+0.0042_-0.0035 ( th . ) , [ eq:1 ] based on a comparison with our next - to - leading order ( nlo ) qcd calculations @xcite .
while this value ( like the one obtained from deep - inelastic electron - photon scattering @xcite ) is in agreement with the current world average of @xmath3 @xcite , it is also less precise , since the latter uses only observables that are known to next - to - next - to - leading order ( nnlo ) of perturbative qcd . in this letter
, we compute the inclusive jet photoproduction cross section for the first time including nnlo contributions , obtained from a unified threshold resummation formalism @xcite , and extract the first nnlo value for the strong coupling constant from photoproduction data .
our calculations are based on our previous work on inclusive jet @xcite and dijet @xcite photoproduction and lead , as we will see , to a considerably improved description of the zeus data and a theoretical error on @xmath4 that is significantly reduced .
the qcd factorization theorem allows to write the differential cross section for inclusive jet photoproduction & = & _ a , b y f_/e(y ) x_f_a/(x _ , _ ) + & & x_p f_b / p(x_p,_p ) _ ab(_s,,_,_p ) as a convolution of the partonic scattering cross section @xmath5 , which includes both direct ( @xmath6 , @xmath7 ) and resolved ( @xmath8 ) photon contributions , with the weizscker - williams flux of photons in electrons @xmath9 and the parton densities in the photon and proton @xmath10 ( @xmath11 for direct photons ) and @xmath12 , respectively . from a unified threshold resummation formalism
a master formula can be obtained that allows to compute soft and virtual corrections to arbitrary partonic hard scattering cross sections @xcite . at nlo
ab&=&_ab^b_s ( ) c_3d_1(z)+c_2d_0(z)+c_1(1-z ) + & + & _ s^d__s+1()a^cd_0(z)+t^c_1(1-z ) , where the second line is only present for processes with complex color flow ( here resolved processes ) , @xmath13 denotes the power of @xmath4 already present in the born term @xmath14 ( 1 for direct and 2 for resolved photoproduction ) , and d_l(z)&=&_+ with decreasing @xmath15 are the leading and subleading logarithms at partonic threshold ( @xmath16 ) in pair - invariant - mass kinematics .
the nnlo master formula is given in the reference cited above , as are the general and complex color flow formul for the coefficients @xmath17 , @xmath18 and @xmath19 .
the coefficients for the simple color flow in direct photoproduction are given here for the first time .
for the qcd compton process
@xmath20 , we find @xmath21 , c_2 & = & c_f--74 + + & + & n_c2 - 1 + -_04 , and the scale - dependent part of the coefficient c_1^&=&-3c_f4 ( ) + ( ) . for the crossed process
@xmath22 , we find @xmath23 , c_2 & = & -5c_f2 + n_c--12 + 12,and the scale - dependent part of the coefficient c_1^&= & -_04 + _ 04 . for both processes , the scale - independent parts of @xmath24 can be found in refs .
these coefficients depend on the qcd color factors @xmath25 and @xmath26 , the one - loop @xmath27-function @xmath28 with @xmath29 quark flavors , the mandelstam variables @xmath30 , @xmath31 and @xmath32 , and the renormalization and proton factorization scales @xmath33 and @xmath34 , but not on the photon factorization scale @xmath35 , as the qed splitting of a photon to a quark - antiquark pair is not enhanced by the logarithms given above .
we therefore only expect the dependencies on the former two scales to be improved at nnlo .
we have verified that at nlo the direct and resolved coefficients agree with the virtual and soft initial - state corrections calculated in pair - invariant - mass kinematics for two jets integrated analytically over singular and numerically over regular regions of phase space @xcite . at nnlo , these coefficients appear exclusively in the three leading logarithmic terms @xmath36 , @xmath37 and @xmath38 of the master formula . for lower terms ,
only the scale - depenent terms are known and included , but the resulting error should be small .
the analytical results described above have been implemented in our program for dijet photoproduction , where a convolution over @xmath39 was already performed for initial - state singularities @xcite .
the nnlo terms for simple color flow were implemented exclusively on the proton side , while those for complex color flow ( resolved photoproduction ) have been split evenly among the photon and the proton . at nlo
, we use of course our complete calculation and not only the logarithmically enhanced terms described above . as a numerical check
, we have repeated the calculation of inclusive jet production at the tevatron at nlo and nnlo with different scales as shown in fig . 2 of ref .
@xcite ( note that the rapidity range there should read @xmath40 ) , finding full agreement ( see our fig .
[ fig:1 ] ) .
relative difference of d0 data , nlo and nnlo predictions with scales @xmath41\times e_t$ ] to the central nlo prediction as a function of jet transverse energy @xmath42 . ]
the final zeus measurements for inclusive jet photoproduction and the determination of @xmath4 were performed with photon virtuality @xmath43 gev@xmath44 , @xmath45 center - of - mass energy in the range @xmath46 gev , and using the inclusive @xmath47-algorithm @xcite with jet radius @xmath48 , transverse energy @xmath49 gev and rapidity in the range @xmath50 @xcite . to facilitate easy comparison of the following figures
, we also employ the zeus - s @xcite and grv - ho @xcite fits of the parton densities in the proton and photon , where the latter has been transformed from the dis@xmath51 to the @xmath52 scheme .
similar measurements have also been performed almost a decade earlier by the h1 collaboration @xcite . due to the lower integrated luminosity available then ( 24.1 pb@xmath0 ) , the data are less precise , but we have verified that within errors they are in good agreement with our calculations at nlo and nnlo .
focusing now on the more precise zeus data , we compare in fig.[fig:2 ] the measured and various theoretical transverse energy relative difference of zeus data , nlo and nnlo predictions with scales @xmath53\times e_t$ ] to the central nlo prediction as a function of jet transverse energy @xmath42 . ]
spectra to the nlo prediction with central scales @xmath54 and after applying hadronization corrections @xcite . the nlo uncertainty band ( blue )
is obtained by varying the scales about the central scale up and down by a factor of two and coincides with the one shown in fig . 2 of ref.@xcite . the nnlo corrections ( including nnlo running of @xmath4 ) increase the central prediction by 3 - 6% , bringing it into considerably better agreement with the experimental data . as expected from the general behavior of threshold logarithms , the increase is larger at high @xmath42 .
the scale uncertainty is also reduced at nnlo ( green band ) , in particular at high @xmath42 , where it drops from 8 to 5% .
note that the data point in the largest @xmath42-bin has been omitted from this comparison , as it suffers from large experimental ( in particular jet energy - scale ) uncertainties . in fig .
[ fig:3 ] we perform a similar comparison for the relative difference of zeus data , nlo and nnlo predictions with scales @xmath53\times e_t$ ] to the central nlo prediction as a function of jet rapidity @xmath55 . ] jet rapidity distribution .
this distribution has been problematic since the earliest hera runs , as it tended to be overestimated in the backward ( photon ) and underestimated in the forward ( proton ) direction .
these discrepancies are indeed observed at nlo in fig .
[ fig:3 ] , together with a large scale uncertainty , in particular in the forward direction .
they were traditionally assigned to hadronization , but also missing higher - order corrections .
this conjecture can now be corroborated for the first time , as the nnlo contributions do bring the theoretical predictions into better agreement with the data in both kinematical regions .
the scale uncertainty is not significantly improved at nnlo , as the data are dominated by the low-@xmath42 region , but within it they can now be described up to the largest rapidities .
note that all figures in this section have been obtained using the world average value for @xmath56 as required for the zeus - s ( standard ) fit @xcite .
to determine the strong coupling constant from these comparisons , the theoretical calculations have to be performed with a set of parton densities in the proton obtained from global fits assuming different values of @xmath57 .
for our analysis at nnlo , we employ the latest fits of the cteq - tea collaboration ( ct10 ) , which have been obtained with nnlo running of the coupling , evolution of the parton densities , deep - inelastic scattering and vector - boson production matrix elements @xcite .
13 different ct10 nnlo sets were used , which correspond to values of @xmath58 to 0.124 .
in contrast , the zeus determination at nlo in eq .
( [ eq:1 ] ) was based on only five different sets of zeus - s parton densities corresponding to values of @xmath59 to 0.123 .
its theoretical error was dominated by higher - order terms , estimated from scale variations , but depended only weakly on variations of the parton densities in the proton , the hadronization model , and the jet algorithm . at nlo , for a scale choice of @xmath60 , and omitting the lowest @xmath42 point , which lies clearly outside the theoretical error band at this order ( see fig .
[ fig:2 ] ) , we reproduce @xmath61 as in eq .
( [ eq:1 ] ) and the zeus analysis @xcite . as stated there ,
including the lowest @xmath42 point severely worsens the quality of the fit , and we find an increase in the minimal value of @xmath62/d.o.f.from 16/6 to 123/7 . through this result
, we also confirm that the fit of the strong coupling constant does not depend very much on the employed parton densities in the proton , which were ct10 nnlo in our fit and zeus - s nlo in the zeus analysis . the uncertainty induced by the parton densities in the photon
was systematically studied by the zeus collaboration and estimated to be @xmath63 % .
it is expected to remain the same at nlo and nnlo , in particular due to the lack of more precise deep - inelastic electron - photon scattering data and a nnlo fit to them . finally , at nnlo and for the same scale choice as the one given above , we obtain _
s(m_z)=0.120^+0.002_-0.002 ( exp.)^+0.003_-0.003(th . ) .
the central value is now lower , since the nnlo contributions increase the cross section for all @xmath42 bins ( see fig .
[ fig:2 ] ) , and is brought closer to the world average of 0.118 .
in addition , the theoretical error is significantly reduced , which reflects the stabilization of the cross section prediction with respect to variations of the unphysical scales .
in conclusion , we have presented here a first calculation of inclusive jet photoproduction up to nnlo of perturbative qcd . leading and subleading logarithmic contributions
were extracted from a unified threshold resummation formalism for photon - parton and parton - parton scattering processes pertinent to direct and resolved photoproduction of jets and shown to agree with those appearing in our full nlo calculations .
the nnlo contributions implemented in our nlo program were shown to correctly reproduce results obtained in the literature for hadron - hadron scattering at the tevatron and to considerably improve the description of final zeus data on jet photoproduction .
a nnlo fit of these data with the ct10 set of parton densities resulted in a new determination of the strong coupling constant at the mass of the @xmath2-boson in agreement with the current world average and the zeus determination at nlo , but with a significantly reduced theoretical error .
we thank c. glasman , n. kidonakis , and w. vogelsang for useful discussions . this work has been supported by the bmbf theorie - verbund `` begleitende theoretische untersuchungen zu den experimenten an den grogerten der teilchenphysik . '' c. adloff _ et al . _
[ h1 collaboration ] , eur .
j. c * 21 * , 33 ( 2001 ) .
s. chekanov _ et al . _
[ zeus collaboration ] , eur .
j. c * 21 * , 443 ( 2001 ) .
f. d. aaron _ et al . _ [ h1 and zeus collaboration ] , jhep * 1001 * , 109 ( 2010 ) .
m. klasen , rev .
phys . * 74 * , 1221 ( 2002 ) .
h. abramowicz _ et al .
_ [ zeus collaboration ] , nucl .
b * 864 * , 1 ( 2012 ) .
m. klasen and g. kramer , z. phys .
c * 72 * , 107 ( 1996 ) and z. phys .
c * 76 * , 67 ( 1997 ) ; m. klasen , t. kleinwort and g. kramer , eur .
j. direct c * 1 * , 1 ( 1998 ) .
s. albino , m. klasen and s. sldner - rembold , phys .
lett . * 89 * , 122004 ( 2002 ) .
j. beringer _
[ particle data group collaboration ] , phys .
d * 86 * , 010001 ( 2012 ) .
n. kidonakis , int . j. mod .
a * 19 * , 1793 ( 2004 ) ; see also d. de florian and w. vogelsang , phys .
d * 76 * , 074031 ( 2007 ) and m. c. kumar and s. o. moch , arxiv:1309.5311 [ hep - ph ] .
m. klasen , g. kramer and s. g. salesch , z. phys .
c * 68 * , 113 ( 1995 ) ; m. klasen and g. kramer , phys .
d * 56 * , 2702 ( 1997 ) .
n. kidonakis and j. f. owens , phys . rev .
d * 63 * , 054019 ( 2001 ) .
s. catani , y. l. dokshitzer , m. h. seymour and b. r. webber , nucl . phys .
b * 406 * , 187 ( 1993 ) ; s. d. ellis and d. e. soper , phys .
d * 48 * , 3160 ( 1993 ) .
s. chekanov _ et al .
_ [ zeus collaboration ] , phys . rev .
d * 67 * , 012007 ( 2003 ) .
m. glck , e. reya and a. vogt , phys .
d * 45 * , 3986 ( 1992 ) and phys . rev .
d * 46 * , 1973 ( 1992 ) . c. adloff
_ et al . _
[ h1 collaboration ] , eur .
j. c * 29 * , 497 ( 2003 ) .
j. gao , m. guzzi , j. huston , h. -l .
lai , z. li , p. nadolsky , j. pumplin and d. stump _ et al .
_ , arxiv:1302.6246 [ hep - ph ] . | we present the first calculation of inclusive jet photoproduction with next - to - next - to - leading order ( nnlo ) contributions , obtained from a unified threshold resummation formalism .
the leading coefficients for direct photoproduction are computed analytically .
together with the coefficients pertinent to parton - parton scattering , they are shown to agree with those appearing in our full next - to - leading order calculations .
for hadron - hadron scattering , numerical agreement is found with a previous calculation of jet production at the tevatron .
we show that the direct and resolved nnlo contributions considerably improve the description of final zeus data on jet photoproduction and that the error on the determination of the strong coupling constant is significantly reduced .
desy 13 - 182 + ms - tp-13 - 34 + |
the decimal digits of @xmath0 are widely believed to behave like as statistically independent random variables taking the values @xmath7 with equal probabilities @xmath8 ( for an overview , see @xcite ) . in section [ powers ]
( it reproduces in part section 2.1 of @xcite ) another similar conjecture is explored - the seemingly almost random behaviour of digits in the base 3 representations of powers @xmath4 .
this conjecture seems to confirm well - it passes even the tests inspired by the central limit theorem and the law of the iterated logarithm .
especially remarkable is fig.[oscilat1 ] below showing oscillations with amplitudes almost as required by the law of the iterated logarithm . in section [ pi ] , similar pictures for the sequences of digits in the decimal representations of the numbers @xmath0 , @xmath5 and @xmath6
are obtained .
the result looks surprising : instead of oscillations with amplitudes required by the law of the iterated logarithm , we observe _ convergence to zero _ !
if , for such `` analytically '' defined irrational numbers , the observed behaviour remains intact _ ad infinitum _
, then the seeming randomness of their digits is only a limited one .
throughout this section , it is assumed that @xmath9 are positive integers such that @xmath10 is irrational , i.e. , @xmath11 for any integers @xmath12 .
let us denote by @xmath13 the * @xmath14-th digit * in the canonical base @xmath15 representation of the number @xmath16 , and by @xmath17 - the * sum of digits * in this representation .
let us consider base @xmath15 representations of powers @xmath18 .
imagine , for a moment , that , for fixed @xmath19 , most of the digits @xmath20 behave like as statistically independent random variables taking the values @xmath21 with equal probabilities @xmath22 .
then , the ( pseudo ) mean value and ( pseudo ) variance of @xmath20 should be @xmath23 the total number of digits in the base @xmath15 representation of @xmath18 is @xmath24 , hence , the ( pseudo ) mean value of the sum of digits @xmath25 should be @xmath26 and , because of the assumed ( pseudo ) independence of digits , its ( pseudo ) variance should be @xmath27 . as the final consequence , the corresponding centered and normed variable @xmath28 should tend to behave as a standard normally distributed random variable with probability density @xmath29 .
one can try to verify this conclusion experimentally .
for example , let us compute @xmath30 for @xmath16 up to @xmath31 , and let us draw the histogram of the corresponding centered and normed variable @xmath32 ( see fig .
[ histo1 ] ) .
as we see , this variable behaves , indeed , almost exactly , as a standard normally distributed random variable ( the solid curve ) . ,
scaledwidth=80.0% ] observing such a phenomenon `` out there '' , one could conjecture that the sum of digits @xmath33 , as a function of @xmath16 , behaves almost as @xmath34 , i.e. , almost * linearly * in @xmath16 . an even more advanced idea for testing randomness of sequences of digits
was proposed in @xcite - let us use the law of the iterated logarithm .
namely , let us try to estimate the amplitude of the possible deviations of @xmath33 from @xmath34 by `` applying '' the law of the iterated logarithm .
let us consider the following centered and normed ( pseudo ) random variables : @xmath35 by summing up these variables for @xmath14 from @xmath36 to @xmath37 , we obtain a sequence of ( pseudo ) random variables : @xmath38 that must obey `` the law of the iterated logarithm .
namely , if the sequence @xmath33 behaves , indeed , as a ' ' typical " sum of equally distributed random variables , then @xmath39 and @xmath40 of the fraction @xmath41 must be @xmath42 and @xmath43 correspondingly ( @xmath44 stands for the natural logarithm ) .
therefore , it seems , we could conjecture that , if @xmath45 then @xmath46 in particular , this would mean that @xmath47 and , for @xmath48 this would mean ( note that @xmath49 ) : @xmath50 @xmath51 @xmath52 , scaledwidth=80.0% ] however , the real behaviour of the expression @xmath53 until @xmath54 does not show convergence of oscillations to the segment @xmath55 $ ] ( see fig . [ oscilat1 ] , obtained by juris erenoks ) .
although @xmath53 is oscillating almost as required by the law of the iterated logarithm , very many of its values lay outside the segment @xmath56 $ ] .
could we hope to prove the above estimate of @xmath33 ?
to my knowledge , the best result on this problem is due to c. l. stewart .
it follows from his theorem 2 in @xcite ( put @xmath57 ) , that @xmath58 where the constant @xmath59 can be effectively computed from @xmath60 .
since then , no better than @xmath61 lower bounds of @xmath33 have been proved .
in section [ powers ] above , the central limit theorem ( fig .
[ histo1 ] ) and the law of the iterated logarithm ( fig .
[ oscilat1 ] ) were used to verify the conjecture that the sum of digits of the base 3 representation of @xmath4 behaves closely to the expected behaviour of the sum of the first @xmath16 members of a sequence of independent random variables taking the values @xmath62 with equal probabilities @xmath63 .
imagine , for a moment , that the digits in the decimal representation of some real number x behave , indeed , like as statistically independent random variables taking the values @xmath7 with equal probabilities @xmath8 .
then , the ( pseudo ) mean value and ( pseudo ) variance of @xmath16-th digit would be ( see the formulas above ) @xmath64 and @xmath65 correspondingly . and , the ( pseudo ) mean value of the sum of the first @xmath16 digits @xmath66 would be @xmath67 , and , because of the assumed ( pseudo ) independence of digits , its ( pseudo ) variance would be @xmath68 .
let us try to estimate the amplitude of the possible deviations of @xmath66 from the expected mean @xmath69 by applying " the law of the iterated logarithm .
let us introduce the necessary centered and normed ( pseudo ) random variables : @xmath70 ( @xmath71 denotes the @xmath14-th digit ) . by summing up these variables for @xmath14 from @xmath36 to @xmath16
, we obtain a sequence of ( pseudo ) random variables : @xmath72 that must obey `` the law of the iterated logarithm .
namely , if the sequence @xmath66 behaves , indeed , as a ' ' typical " sum of equally distributed random variables , then @xmath39 and @xmath40 of the fraction @xmath73 must be @xmath42 and @xmath43 correspondingly ( @xmath44 stands for the natural logarithm )
. however , fig .
[ oscilatpi ] , fig .
[ oscilate ] , and fig .
[ oscilatsqrt2 ] obtained by juris erenoks for the first @xmath76 digits of the numbers @xmath0 , @xmath5 and @xmath6 are showing a completely different behaviour .
( digits were provided by wolfram mathematica @xcite . ) for the three numbers in question , the values of @xmath75 do not oscillate accross the entire segment @xmath55 $ ] , instead , they seem converging to 0 . thus , the pictures seem to support the following somewhat stronger conjecture for @xmath0 , @xmath5 and @xmath6 : @xmath77 , iraids j. , opmanis m. , opmanis r. , podnieks k. : integer complexity : experimental and analytical results ii .
http://arxiv.org/abs/1409.0446[arxiv:1409.0446 ] (
september 2014 ) [ last accessed : 13 november 2014 ] | the decimal digits of @xmath0 are widely believed to behave like as statistically independent random variables taking the values @xmath1 , @xmath2 with equal probabilities @xmath3 . in this article , first , another similar conjecture is explored - the seemingly almost random behaviour of digits in the base 3 representations of powers @xmath4 .
this conjecture seems to confirm well - it passes even the tests inspired by the central limit theorem and the law of the iterated logarithm .
after this , a similar testing of the sequences of digits in the decimal representations of the numbers @xmath0 , @xmath5 and @xmath6 was performed .
the result looks surprising : unlike the digits in the base 3 representations of @xmath4 , instead of oscillations with amplitudes required by the law of the iterated logarithm , _ convergence to zero _ is observed .
if , for such `` analytically '' defined irrational numbers , the observed behaviour remains intact _
ad infinitum _
, then the seeming randomness of their digits is only a limited one . |
in recent years , the study of meteors received significant input primarily due to the campaigns to observe the leonid meteor storms ( jenniskens et al . 2000 ) .
the new studies introduced novel observational methods as well as new analysis techniques . among those , the analysis of the light curves of visible meteors remains one of the more widespread techniques .
the accepted picture regarding the light production by a meteor is that collisions between the ablated meteor atoms and atmospheric molecules and ions are responsible for this phenomenon . for small meteoroids ,
the light production is proportional to the loss of kinetic energy by the body .
as the meteoroid velocity remains almost constant throughout the luminous phase , it follows that the light production tracks the instantaneous mass loss by ablation .
a general review of meteor light curve analysis was presented by hawkes et al .
they discussed six improvements to traditional meteor light curve analysis in order to provide higher resolution and better information : use of generation iii image intensifier technology , digital recording techniques , digital image processing algorithms which separate the even and odd video fields , whole - image background subtraction , analysis of pixel - by - pixel high resolution meteor light curves , and utilization of coincidence and correlation techniques .
however , they did not introduce novel analysis methods of the higher quality data .
the classical light curve ( lc ) produced by a solid , compact , and non - fragmenting meteoroid should be smooth and exhibit its maximum luminosity near the end of the trail ( cook 1954 ) .
this is the combination of the exponential increase in air density as the meteoroid penetrates into deeper atmospheric layers , and of the reduction in the surface area presented by the meteoroid to the airflow as the ablation proceeds .
however , recent measurements indicate that this picture of single - body ablation may not be correct . in many cases ,
faint meteors were shown to produce @xmath2symmetric lcs . a basic question concerning the behavior of the light produced by ablating meteors is therefore whether this process involves at all stages a single object , or whether during the production of light the meteoroid disintegrates in a rather large number of grains ( the `` dustball '' model : hawkes & jones 1975 ) .
this model assumes that the meteoroid is composed of numerous small grains with a high melting point temperature , held together by a low melting point glue .
fisher et al .
( 2000 ) followed hawkes & jones ( 1975 ) and suggested that most meteoroids are collections of hundreds to thousands of fundamental minute grains , at least some of which are released prior to the onset of intensive ablation .
one would expect these grains , unless extremely uniform in physical properties , to become aerodynamically separated during atmospheric entry , and therefore to produce a `` wake '' , which is defined as some instantaneous meteor light production from an extended spatial region .
fisher et al . presented theoretical results for wake production as a function of grain mass distribution , height of separation , zenith angle and velocity .
koten & borovicka ( 2001 ) analyzed 234 meteor lcs , among which there were 110 leonids from the 1998 and 1999 showers .
one of their goals was the identification of relations between the lc shapes and other parameters . among these , they included the leading and the trailing slopes of the lcs , defined as linear relations between the beginning of the lc and its maximum , and between the maximum and the terminal point of the lc .
bellot rubio et al ( 2002 ) studied photographic light curves of relatively bright meteors with magnitudes in the range from + 2.5 to -5 , collected by jacchia et al .
( 1967 ) with super schmidt observations , in order to derive the average density of meteors and to test whether the single - body theory of meteor evolution fits the observations better than that of continuous disintegration .
velocities , decelerations and magnitudes were fitted simultaneously to synthetic light curves , and the ablation coefficient , the shape - density coefficient and the pre - atmospheric mass of each individual meteoroid were determined .
bellot rubbio et al . could not confirm the large meteor density values determined from the quasi - continuous fragmentation models , essentially supporting the single - body ablation model .
babadzhanov ( 2002 ) reached opposite conclusions , supporting the continuous fragmentation model , from an analysis of 111 photographic light curves of meteors .
similar conclusions , that is a preference for continuous disintegration during re - entry of meteors , were reached by jiang & hu ( 2001 ) from an analysis of high spatial resolution meteor light curves obtained during the 1998 and 1999 leonid showers .
light curves of leonid meteors collected in recent years were analyzed by murray et al .
the meteors they concentrated on were fainter than those discussed by bellot rubbio et al .
( 2002 ) , of 6 - 8 mag , and the observations were performed with intensified ccd video cameras with fields of view from 16@xmath312@xmath4 to 40@xmath335@xmath4 .
murray et al .
concentrated on the modified _
f _ parameters and used these to distinguish among differences in overall light curve shapes .
the _ f _ parameter , to be explained below , ranged between 0.49 and 0.66 for individual annual showers ( where _ _ f__=0.5 is defined as a symmetrical curve , _
_ [email protected] an early skewed one , and _ _
[email protected] a late skewed curve ) .
the findings indicate morphological differences between leonid meteoroids observed in different years , thus originating from different ejection epochs .
murray et al . also noted the presence of very distinctive ( but not quantified ) features among yearly events , with the 1998 leonid light curves characterized by early skewed shapes while those from 1999 showed unusual `` flat topped '' curves .
a similar conclusion was reached by koten & borovicka ( 2001 ) . in this , and in all the previously mentioned publications , the lcs were discussed almost exclusively in reference to the single quantitative parameter @xmath0 , the skewness of the lc ( fleming et al .
1993 ) .
the ungluing of a complex of micro - meteoroids , with the subsequent formation of a classical trail , the epitome of the fragmentation model , was shown to fit the vhf radar observations of geminid meteors ( campbell - brown & jones 2003 ) .
the interesting finding of this study is the altitude at which the radius of the trail is zero , and which can be interpreted as the region where the ungluing process begins : approximately 240-km for the geminids .
this fits the region where the high altitude radar echoes connected with meteor activity were detected by the israeli l - band radar ( brosch et al .
2001 ) .
in addition , the light produced by many meteors was noted to be far from `` well - behaved '' and steady .
smith ( 1954 ) computed a simple model to account for the sudden brightening of meteor light curves .
these flares cause the brightening of the meteor by 1 - 2 mag and , as explained by smith , are the result of the release of a few thousand particles from the original single - body meteor .
one explanation for this phenomenon , put forward by kramer & gorbanev ( 1990 ) , is that this flaring is produced by the shedding and spraying of a liquid film formed on the leading surface of the ablating meteor .
they described a phenomenon by which the light intensity from the meteor is @xmath2constant , then shows a sudden drop ( depression ) , after which the meteor flares in brightness . therefore , meteor lcs may show both sudden brightening episodes , as well as sudden dimmings .
kramer & gorbanev remarked that meteors showing a depression in their lc have the brightness maximum earlier , on average , than meteors with no depression . in terms of the skewness parameter , such meteors would then be classified as `` early skewers '' .
it seems that in order to discuss statistical properties of meteors it is necessary to use descriptors that would reduce the amount of data characterizing a single lc while providing quantitative measures .
the descriptors could be the symmetry parameter @xmath0 described above and used extensively , or the leading and trailing slopes as described by murray et al .
( 1999 ) and by koten & borovicka ( 2001 )
. these could , in principle , refine the studies where tens to hundreds of lcs are collected and simultaneously analyzed , by reducing each lc to a small set of consistent numbers .
the present paper presents an exploration of measurable parameters in the context of the examination of meteor lcs collected in israel during the 2002 leonid shower . a first attempt was already made in our contribution describing lcs of leonids 2001 , geminids 2001 , and perseids 2002 ( brosch et al .
2002 ) , where the pointedness parameter
_ p _ was introduced ( see below ) .
the 2002 leonid shower was analyzed by arlt et al .
( 2002 ) with the following preliminary characteristics .
the activity was due to two dust trails , one of cometary dust ejected seven revolutions ago in 1767 that produced a peak zhr of 2510@xmath760 and was seen in asia and europe , and the other by dust ejected four revolutions ago in 1866 that produced a zhr of 2940@xmath7210 and was seen in the americas .
the first peak took place on november 19 at 04:10@xmath71 ut with a full - width at half - maximum ( fwhm ) of 39@xmath73 minutes .
the second was in the same day at 10:47@xmath71 ut with a fwhm of 25@xmath73 minutes .
our observations described below covered the period of the rise toward the twin peaks and a portion of the decay from the peak activity , but neither of the peaks themselves . here
we expand the discussion to include more parameters and to explore the internal correlations they might show for the case of the meteors observed in november 2002 during the leonid shower by using classical statistical techniques .
this approach is novel in the study of meteor lcs and may prove useful in uncovering hidden connections among the measured parameters , as many lcs are collected and uniformly analyzed in this manner .
we describe our observations in section 2 , the data reduction in section 3 , the analysis and results in section 4 , and conclude in section 5 .
starting in november 1998 , the wise observatory ( wo ) is active in the observation of meteors . from 2001 onward ,
these observations consist of intensified video ( iccd ) measurements , sometimes accompanied by l - band phased - array radar observations .
we do not discuss here the radar observations ( e.g. , brosch et al .
2001 ) , but concentrate exclusively on the analysis of the light curves ( lcs ) derived from the intensified video observations .
the observations reported and analyzed here were collected during five nights , from november 15 - 16 to november 19 - 20 2002 , using mobile meteor detection systems .
each system is based on an itt night vision 18-mm , iii - generation image intensifier ( it ) with a gaas cathode , supplied by collins electro - optics .
the it is optically - coupled to an astrovid 2000 ccd video camera operating in the pal tv standard ( 50 interlaced half - frames per second , or a 20 msec exposure for each half - frame ) .
the it is illuminated by a 50-mm f/0.95 navitar lens and provides a final imaged field of 6@xmath88 degrees@xmath9 .
the astrovid 2000 sends the video stream to a digital hi8 video recorder equipped with a date / time stamper and , in parallel , to a matrox meteor ii frame grabber card mounted in the docking bay of a compaq armada 850 computer .
the log of observations is given in table 1 and shows the number of meteors recorded by the observing stations during each individual night as well as the total number of detected meteors .
.meteor observations , by night and by camera [ cols="<,^,^,^",options="header " , ] note that beech & murray ( 2003 ) calculated also the effect of the zenith angle on the light curve .
their figure 7 displays this effect for the @xmath10=1.73 model with the following resuts : z=60@xmath4 ; _ _ f__=0.48 ; z=45@xmath4 : _ _ f__=0.45 , _ _ p__=0.87 ; and z=0@xmath4 : _ _ f__=0.45 , _
_ p__=0.70 .
as mentioned above , their lcs are cut off at high altitude presumably by the evolution beginning at 120 km ; the effect is primarily causing high values for _ p _ and the true _
p _ values can not be easily recovered .
we presented an objective exploration of some measurable parameters of meteor light curves based on a uniform collection of lc from the 2002 leonid shower .
the lcs belong mostly to leonid and sporadic meteors and this allowed an inter - comparison of the properties of these two groups .
the analysis confirmed the reduced importance of the symmetry parameter @xmath0 used in many previous lc analyses , and showed that , at least for te meteor lcs studied here , the most important parameters are the duration , the skewness , and the brightness amplitude . comparisons of the distributions of measured parameters between leonid and sporadic meteors indicated that the leo lcs are more uniform than the spo ones .
the statistical analysis showed that short - duration meteors have stronger sudden brightening and/or dimming episodes ( flares ) than long - duration ones .
the long duration meteors have shallower general brightening and dimming stages .
light curves with early skews , that reach their maximal brightness before the middle of their luminous phase , tend to have lcs with steep slopes .
a pca showed that , at least for the meteors observed in 2002 , three principal components suffice to describe most of the variation observed among the lcs .
the components are related ( in order of importance ) to the duration of a trail , its brightness amplitude , and its amount of skewness .
the `` flaring '' activity presented by a meteor train seems to be related both to the duration of the trail and to its brightness amplitude .
an attempt to compare the meteor lcs characterized here with theoretical models was not very successful because of a lack of published model light curves and of characterizing parameters , such as those defined and tested here .
however , the use of both _ f _ and _ p _ in comparing the observations with the models of murray et al .
( 2000 ) indicated similar population indices as estimated for the 2002 leonids by arlt et al ( 2002 ) .
a comparison with the more recent models of beech & murray ( 2003 ) yielded a population index of @xmath11 , rather different from that measured for the 2002 leonids , but probably affected by the lack of data describing the high altitude lc behaviour .
we acknowledge support from the israel science foundation to study meteors in israel .
discussions with mr .
eran ofek and mr .
ilan manulis , and the continuing help of the wise observatory staff in securing meteor observations , are appreciated .
this paper has greatly benefited from the remarks of an anonymous referee .
brosch , n. , helled , ravit , polishook , d. , schijvarg , s. & manulis , i. 2002 , in `` asteroids , comets , meteors - acm 2002 '' .
( barbara warmbein , ed . ) esa sp-500 .
noordwijk : esa publications division , p. 209
fleming , b. d. e. , hawkes , l. r. & jones , j. 1993 , in `` meteoroids and their parent bodies '' ( j. stohl and i.p .
williams , eds . ) , bratislava : astronomical institute , slovak academy of sciences , 1993 , p.261 murray , i.s . , beech , m. , taylor , m.j . , jenniskens , p. & hawkes , r.l .
2000 , in `` leonids storm research '' ( p. jenniskens , f. rietmeijer , n. brosch , and p. fonda , eds . ) , dordrecht : kluwer academic publishers , p. 351 | we investigate a uniform sample of 113 light curves of meteors collected at the wise observatory in november 2002 during a campaign to observe the leonid meteor shower .
we use previously defined descriptors , such as the classical skewness parameter @xmath0 and a recently - defined pointedness variable @xmath1 , along with a number of other measurable or derived quantities , in order to explore the parameter space in search of meaningful light curve descriptors . in comparison with previous publications ,
we make extensive use of statistical techniques to reveal links among the various parameters and to understand their relative importance . in particular , we show that meteors with long - duration trails rise slowly to their maximal brightness and also decay slowly from the peak , while showing milder flaring than other meteors .
early skewed meteors , with their peak brightness in the first half of the light curve , show a fast rise to the peak . we show that the duration of the luminous phase of the meteor is the most important variable differentiating among the 2002 meteor trails .
the skewness parameter @xmath0 , which is widely used in meteor light curve analyses , appears only as the second or third in order of importance in explaining the variance among the observed light curves , with the most important parameter being related to the duration of the meteor light - producing phase .
we suggest that the pointedness parameter @xmath1 could possibly be useful in describing differences among meteor showers , perhaps by being related to the different compositions of meteoroids , and also in comparing observations to model light curves .
we compare the derived characteristics of the 2002 meteors with model predictions and conclude that more work is required to define a consistent set of measurable and derived light curve parameters that would characterize the light production from meteors , and suggest that meteor observers should consider publishing more characterizing parameters from the light curves they collect .
theorists describing the light production from meteors should present their results in a form better compatible with observations .
meteors , light curves , statistical analysis |
let @xmath2 denote the @xmath4-dimensional euclidean space with an orthonormal basis @xmath5 . for convenience
we use small letters to denote real numbers , use small bold letters to denote points ( or vectors ) and use capital letters to denote sets of points in the space . in particular , let @xmath6 denote the origin of @xmath2 , and let @xmath7 denote the @xmath4-dimensional unit ball @xmath8 .
a metric @xmath9 define in @xmath2 is called a _ minkowski metric _ if it satisfies the following conditions . * 1 . *
_ @xmath10 holds for all vectors @xmath11 and the equality holds if and only if @xmath12 _ * 2 . * _ @xmath13 holds for all vectors @xmath11 and all real numbers @xmath14 . _
_ @xmath15 holds for all vectors @xmath16 @xmath17 .
_ let us define @xmath18 since @xmath19 , if @xmath20 , then @xmath21 .
therefore @xmath3 is centrally symmetric . on the other hand , if both @xmath22 and @xmath23 belong to @xmath3 and if @xmath24 is a real number satisfying @xmath25 , by the third condition of the metric we have @xmath26 in other words , @xmath3 is convex . in addition
, the origin @xmath6 is an interior point of @xmath3 .
otherwise there would be a point @xmath22 satisfying @xmath27 . finally , since @xmath28 holds if and only if @xmath29 , it can be deduced that @xmath3 is bounded . as a conclusion , @xmath3 is a centrally symmetric convex body centered at the origin @xmath6 .
on the other hand , each centrally symmetric convex body @xmath3 centered at the origin @xmath6 can define a minkowski metric .
let @xmath30 denote the point in the direction of @xmath22 and on the boundary of @xmath3 , and define @xmath31 in other words , @xmath32 is the positive number @xmath33 that @xmath22 is on the boundary of @xmath34 .
it can be easily verified that @xmath35 holds for all @xmath36 , and @xmath37 holds for all vectors @xmath11 and all real numbers @xmath14 . since @xmath3 is convex , for any pair of vectors @xmath22 and @xmath23 it follows that @xmath38 @xmath39 and therefore @xmath40 thus @xmath9 defines a minkowski metric on @xmath2 . in particular , when @xmath3 is the @xmath4-dimensional unit ball , the corresponding minkowski metric is the euclidean metric . as a conclusion of this section , we get the following well - known assertion ( see page 7 of @xcite ) . * theorem 0 . *
_ there is a one - to - one correspondence between minkowski metrics and centrally symmetric convex bodies in @xmath2 . _
let @xmath41 denote a minkowski metric determined by a centrally symmetric convex body @xmath3 and let @xmath42 denote the _
minkowski distance _ between two points @xmath22 and @xmath23 with respect to the metric . for two distinct points
@xmath43 and @xmath44 in @xmath2 we define @xmath45 and @xmath46 where @xmath47 . in other words , @xmath48 is a hyperplane containing the origin and perpendicular to the vector @xmath49 , and @xmath50 is the straight line determined by the point @xmath22 and the vector @xmath51 .
then , for every @xmath47 , we define @xmath52 now let us introduce a basic result about @xmath53 which is essential for the definition of the minkowski bisectors .
* lemma 1 ( horvth @xcite ) . * _ for every @xmath54 , the set @xmath55 is either a single point or a closed segment .
_ * proof . *
since @xmath3 is a centrally symmetric convex body , both @xmath56 and @xmath57 are continuous functions of @xmath23 . on the other hand ,
when @xmath14 is sufficiently large , we have @xmath58 and @xmath59 thus it follows that @xmath55 is nonempty .
in addition , it can be easily deduced that @xmath60 is always a compact set .
now we show that if both @xmath61 and @xmath62 belong to @xmath63 ( see figure 1 ) then the whole segment @xmath64 $ ] belongs to @xmath55 .
assume that @xmath65 @xmath66 and @xmath67 .
it is easy to see that both @xmath68 and @xmath69 belong to @xmath70 , @xmath61 is between @xmath62 and @xmath71 and therefore @xmath72 holds for some @xmath73 with @xmath25 .
thus we get @xmath74 by this contradiction we can conclude that @xmath75
. then it follows by convexity ( since all @xmath71 , @xmath69 , @xmath61 and @xmath62 belong to the boundary of @xmath76 ) that the whole segment @xmath77 $ ] belongs to the boundary of @xmath78 , the whole segment @xmath79 $ ] belongs to the boundary of @xmath80 , and therefore @xmath64 $ ] belongs to @xmath55 .
* definition 1 . * _ let @xmath43 and @xmath44 be fixed distinct points
. for @xmath47 let @xmath81 denote the middle point of @xmath55 .
then we define @xmath82 and call it a minkowski bisector of @xmath83 with respect to @xmath41 .
_ * remark 1 . * _ in the literature like @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and @xcite , the bisector was defined by @xmath84 clearly we have @xmath85 as one will see from section @xmath86 that , for the purpose to study packing and covering , @xmath87 is more natural than @xmath88 .
_ let @xmath89 be a map from @xmath48 to @xmath90 defined by @xmath91 .
clearly it is a one - to - one map .
however it is less obvious if it is continuous in general .
we note that in some references such as @xcite and @xcite our continuous `` is referred as homeomorphic to a hyperplane '' .
in fact , as one can see from the following lemma and example , the situation is quite complicated .
* lemma 2 . * _ in @xmath92 , for any given metric @xmath93 and given distinct points @xmath43 and @xmath44 , the map @xmath94 from @xmath48 to @xmath90 is continuous . _
* proof . * without loss of generality , we assume that @xmath95 and @xmath96 . then the set @xmath97 is the @xmath98-axis .
let @xmath99 denote the set of points @xmath100 such that @xmath53 is a single point and write @xmath101 . by repeating partial argument of lemma 1
, it can be deduced that @xmath99 is closed and if @xmath102 , then @xmath103 whenever @xmath104 . in addition , if @xmath105 , @xmath106 and @xmath107 , then @xmath89 ( for @xmath108 and @xmath109 ) is the middle point of @xmath110 $ ] , where @xmath111 and @xmath112 , as shown in figure 2 . in other words
@xmath89 is an half straight line for @xmath113 .
similarly , one can deal with the case that @xmath108 and @xmath114 .
then the continuity of @xmath89 follows easily , by considering two cases @xmath115 and @xmath116 .
* example 1 . *
_ in @xmath0 let us define @xmath117 where @xmath118 and @xmath119 .
it is easy to see that @xmath3 is a centrally symmetric convex body and its surface has only two vertical segments , they are @xmath120 $ ] and @xmath121 $ ]
. take @xmath122 and @xmath123 and let @xmath124 denote the set of all planes which containing both @xmath43 and @xmath44 . by considering the intersections with planes @xmath125 , one can deduce that the minkowski bisector @xmath87 with respect to @xmath41 consists of two parts @xmath126 and @xmath127 , where @xmath126 is a straight line @xmath128 and @xmath127 is a set between two planes @xmath129 and @xmath130 , as shown in figure @xmath86 .
therefore it is easy to see that the map @xmath91 is not continuous at the points @xmath131 whenever @xmath132 . by adding more edges similar to @xmath120 $ ]
, one can make the situation much more complicated .
_ let @xmath133 denote the set of all @xmath4-dimensional centrally symmetric convex bodies centered at the origin @xmath6 and let @xmath134 denote the hausdorff metric defined on @xmath133 . in other words ,
@xmath135 let @xmath43 and @xmath44 be two fixed distinct points , let @xmath136 be a centrally symmetric convex body , and let @xmath137 , @xmath138 , @xmath139 be a sequence of centrally symmetric convex bodies .
it is natural to ask , would @xmath140 converges to @xmath141 if @xmath142 unfortunately , as one will see from the next example , the answer to this question is negative , even in the plane .
* example 2 .
* _ in the plane we define @xmath143 and @xmath144 for @xmath145 then we have @xmath146 let @xmath147 and @xmath148 , we have @xmath149 however , as shown by figure @xmath150 , the minkowski bisector @xmath140 consists of points @xmath151 satisfying @xmath152 therefore the sequence @xmath140 does not converge to @xmath153 _
if @xmath87 is a continuous surface , then it divides the whole space nicely into two half spaces .
unfortunately , as it was shown by example 1 , the minkowski bisectors are not always continuous .
nevertheless , for many important metrics , the continuity can be guaranteed by the next lemma .
* lemma 3 . * _ if @xmath3 is a regular centrally symmetric convex body or a centrally symmetric polytope , for any pair of distinct points @xmath43 and @xmath44 , the map @xmath154 from @xmath155 to @xmath87 is continuous . _ * proof . *
first let us deal with the regular case . by regular we mean that its surface does not contain any segment .
then , for any point @xmath47 , @xmath156 is a single point .
if the map is not continuous at point @xmath157 , then one can deduce that @xmath158 is not a single point .
the regular case follows .
now we consider the polytope case .
let @xmath159 be an @xmath4-dimensional centrally symmetric convex polytope , let @xmath160 denote its projection on @xmath48 , and let @xmath161 denote the relative boundary of @xmath160 .
clearly @xmath160 is a @xmath162-dimensional centrally symmetric polytope and @xmath161 is a polytope complex of dimension less than or equal to @xmath163 .
let @xmath62 be a point in @xmath161 , let @xmath164 denote the point @xmath165 with maximal @xmath73 , let @xmath166 denote the corresponding point with minimal @xmath73 , and write @xmath167 since @xmath159 is a polytope , by convexity , it is easy to see that both @xmath168 and @xmath169 are continuous maps for @xmath170 , and @xmath171 is a continuous function for @xmath170 .
let @xmath172 denote the two - dimensional hyperplane determined by @xmath43 , @xmath44 and @xmath62 .
by studying the intersections with @xmath172 , it can be shown that @xmath173 is a single point if and only if @xmath174 ( when @xmath175 , @xmath176 can be any number ) . for @xmath177
, let @xmath178 denote the point @xmath179 in @xmath173 with the maximal @xmath24 and let @xmath180 denote the corresponding point in @xmath181 with the minimal @xmath73 . when @xmath182 , illustrated by figure 5 , it can be shown by similar triangles that @xmath183 @xmath184 and @xmath185 let @xmath186 denote the set of the points @xmath100 such that @xmath187 is not a single point .
it can be shown that @xmath188 is an open set and therefore @xmath189 is a closed set . by ( 1 ) it follows that @xmath89 is continuous in @xmath190 . on the other hand @xmath89 is continuous in @xmath191 .
therefore @xmath89 is a continuous map from @xmath48 to @xmath192 .
the lemma is proved . by the previous proof , it is easy to see that the minkowski bisector @xmath193 is consists of planar pieces if the metric is defined by a polytope @xmath159
. let @xmath194 denote the minimal number @xmath195 such that the minkowski bisector of @xmath43 and @xmath44 with respect to @xmath196 can be divided into @xmath195 planar pieces .
we have the following results
. * theorem 1 . * _ if @xmath159 is a centrally symmetric polygon with @xmath197 vertices .
then , we have _ @xmath198 * proof . * for convenience , without loss of generality , we take @xmath147 and @xmath199
. let @xmath33 change from @xmath200 to infinity , it is easy to see that the boundaries of @xmath201 and @xmath202 intersect each other whenever @xmath203 for some @xmath176 .
for some @xmath204 , assume that the boundaries of @xmath205 and @xmath206 meet at a vertex @xmath207 of @xmath208 or @xmath206 ( see figure 6 ) , @xmath209 is an edge of @xmath210 , @xmath211 is an edge of @xmath212 .
let @xmath213 denote the line which is parallel to @xmath209 and pass through @xmath214 , let @xmath215 denote the line which is parallel to @xmath211 and pass through @xmath216 , let @xmath217 be the intersection of @xmath213 and @xmath215 , let @xmath126 denote the line passes @xmath207 and @xmath217 , let @xmath218 denote the line determined by @xmath43 and @xmath219 , and let @xmath220 denote the line determined by @xmath44 and @xmath221 .
then @xmath218 intersects @xmath126 at a point @xmath222 and @xmath220 intersects @xmath126 at a point @xmath223 .
let @xmath61 denote the @xmath224 which is closer to @xmath207 . by elementary geometry
it is easy to see that the whole segment @xmath225 $ ] belongs to @xmath193 .
in addition , for some @xmath226 , the boundaries of @xmath227 and @xmath228 meet at @xmath61 and @xmath61 is a vertex of either @xmath229 or @xmath228 . by repeating this process and dealing with two cases with respect to if @xmath230 contains a segment which is parallel to @xmath231 or not , it can be shown that @xmath198 the theorem is proved . * remark 2 .
* _ it is easy to see that @xmath232 holds for all positive number @xmath14 .
thus , for a fixed polytope @xmath159 , the number @xmath194 is determined by the direction of @xmath231 .
in fact , it follows by the proof of theorem @xmath233 that , @xmath234 holds for all @xmath43 and @xmath44 , except for a finite number of directions @xmath231 . _ in higher dimensions the situation is much more complicated .
we have the following upper bound for @xmath194
. * theorem 2 . * _ let @xmath235 be an integer and let @xmath159 be an @xmath4-dimensional centrally symmetric polytope with @xmath197 facets
. then @xmath236 holds for any pair of distinct points @xmath43 and @xmath44 .
_ * proof .
* let @xmath237 denote a facet of @xmath159 , let @xmath238 denote its outer unit normal , and let @xmath239 denote the set of the @xmath197 facets of @xmath159 .
then we define @xmath240 @xmath241 and @xmath242 it is easy to see , let @xmath243 denote the number of the elements of @xmath244 , that these three sets are pairwise disjoint and satisfying @xmath245 and @xmath246 assume that @xmath247 can be divided into @xmath248 planar pieces @xmath249 , @xmath250 , @xmath139 , @xmath251 and write @xmath252 for each @xmath253 let @xmath254 be a relative interior point of @xmath255 such that there are a relative interior point @xmath224 of a facet which belongs to either @xmath256 or @xmath257 , a relative interior point @xmath258 of a facet which belongs to either @xmath257 or @xmath259 , and a positive number @xmath260 satisfying @xmath261 if @xmath262 , by looking at figure 5 , it can be deduced that @xmath263
. then @xmath264 can be divided into two disjoint subsets @xmath265 and @xmath266 clearly we have @xmath267 now we proceed to estimate @xmath268 and @xmath269 , respectively .
if @xmath224 and @xmath270 belong to one facet @xmath271 in @xmath256 and @xmath258 and @xmath272 belong to one facet @xmath273 in @xmath259 , then one can deduce that @xmath254 and @xmath274 should belong to the same planar piece .
in other words , the whole segment @xmath275 $ ] belongs to @xmath193 . to see this , when @xmath276 , writing @xmath277 and @xmath278 , we have @xmath279 and @xmath280 where @xmath281 therefore @xmath268 is bounded from above by the number of distinct pairs of facets @xmath282 such that @xmath283 and @xmath284 .
then by ( 2 ) we get @xmath285 for @xmath286 we write @xmath287 let @xmath288 denote the set of the facets @xmath273 of @xmath159 such that @xmath289 and @xmath290 is a @xmath291-dimensional polytope and let @xmath292 denote the set of the facets @xmath271 of @xmath159 such that @xmath293 and @xmath294 is a @xmath291-dimensional polytope
. for @xmath224 , @xmath258 and @xmath254 defined by ( 4 ) , let @xmath295 denote the point @xmath296 with the maximal @xmath14 , and let @xmath297 denote the point @xmath296 with the minimal @xmath14 .
then , we have @xmath298 and @xmath299 .
if both @xmath295 and @xmath300 belong to @xmath301 for a facet @xmath302 and both @xmath303 and @xmath304 belong to @xmath305 for a facet @xmath306 , by an argument similar to the previous case it can be deduced that both @xmath254 and @xmath274 belong to a planar piece of @xmath247 . therefore @xmath307 is bounded from above by the number of distinct pairs of facets @xmath308 such that @xmath302 and @xmath306 .
consequently , we get @xmath309 and , by ( 3 ) , @xmath310 as a conclusion of ( 5 ) , ( 6 ) and ( 7 ) we get @xmath311 the theorem is proved . * remark 3 .
* _ although we can not give a proof , the upper bound in theorem @xmath312 seems much too large .
it is easy to see that the direction @xmath231 such that @xmath313 is a zero measure set on @xmath314 .
therefore we have @xmath315 unless the direction @xmath83 belongs to a zero measure set of @xmath314 . _
let @xmath3 be an @xmath4-dimensional centrally symmetric convex body , and let @xmath43 and @xmath44 be two distinct points in @xmath2 .
we recall that @xmath48 is the hyperplane @xmath316 and @xmath30 is the point @xmath317 in @xmath87 . for convenience ,
we define @xmath318 in principle , the structure of @xmath319 can be very complicated .
nevertheless , we have the following general result . on the contrary , if @xmath328 holds for some @xmath329 , we proceed to deduce a contradiction .
assume that @xmath330 with suitable @xmath331 and @xmath332 then the point @xmath333 corresponding to @xmath23 on the minkowski bisector is below @xmath62 , as shown in figure 7 .
by lemma 2 , there is a suitable @xmath334 between @xmath43 and @xmath62 such that @xmath335 .
therefore , @xmath336 holds for some suitable @xmath14 .
then we have @xmath337 if the whole segment
@xmath338 $ ] belongs to @xmath339 , then one can deduce that the whole segment @xmath340 $ ] belongs to @xmath341 and thus @xmath342 , which contradicts the assumption .
if @xmath343\not\subset\partial ( \lambda c+{\bf q}),\ ] ] then by convexity it can be deduced that @xmath344 which contradicts ( 8) . * definition 2 . * _ let @xmath3 be a centrally symmetric convex body , let @xmath347 be a lattice and define @xmath348 we call @xmath349 a minkowski cell of @xmath347 with respect to the metric @xmath350 . _ * remark 5 . *
_ for fixed @xmath3 , @xmath43 and @xmath44 , it can be verified that both @xmath232 and @xmath351 hold for all positive numbers @xmath14 .
if @xmath347 is a lattice and @xmath352 is a non - singular linear transformation from @xmath2 to @xmath2 .
then we have @xmath353 @xmath354 and @xmath355 by lemma @xmath150 it follows that the minkowski cells are centrally symmetric star sets centered at the origin . in the euclidean case , @xmath356 , all minkowski bisectors are hyperplanes , and all minkowski cells are parallelohedra . _ * lemma 5 . *
_ let @xmath3 be an @xmath4-dimensional centrally symmetric convex body , let @xmath347 be an @xmath4-dimensional lattice , and let @xmath357 denote the smallest positive number such that @xmath358 is a covering of @xmath2 . then _
@xmath359 * proof . *
if , on the contrary , there is a point @xmath360 such that @xmath361 since @xmath362 , there is a lattice point @xmath363 such that @xmath364 thus , we have @xmath365 and hence @xmath366 the lemma is proved .
* theorem 3 . *
_ let @xmath3 be a two - dimensional centrally symmetric convex domain and let @xmath347 be a two - dimensional lattice . then the region @xmath349 is a centrally symmetric star domain and @xmath367 is a tiling . _
if , on the contrary , without loss of generality there are a point @xmath22 , a positive number @xmath369 and a lattice point @xmath62 such that @xmath370 then , we observe the minkowski bisector @xmath371 . if @xmath372 , then we have @xmath373 and @xmath374 which contradicts to ( 9 ) , where @xmath375 is the boundary point of @xmath3 in the direction of @xmath62 as defined in section 1 . if @xmath376 , since @xmath371 divides @xmath92 into two separated parts which contains @xmath377 and @xmath378 , respectively .
therefore , @xmath379 ca nt belong to both @xmath377 and @xmath378 simultaneously when @xmath369 is sufficiently small , which contradicts to ( 9 ) as well . without loss of generality , since @xmath347 is periodic , we assume that @xmath382 . clearly , we have @xmath383 whenever @xmath384 therefore , there are at most @xmath385 minkowski bisectors which can effect @xmath349 .
consequently , we assume further that @xmath386 if , for any positive @xmath369 there is a point @xmath387 such that @xmath388 and @xmath389 holds for all @xmath390 then , since @xmath349 is compact , we have @xmath360 . if , there are a positive number @xmath391 and a subset @xmath392 of @xmath393 with @xmath394 such that @xmath395 holds for all @xmath396 and @xmath397 .
then , @xmath398 holds for some suitable positive number @xmath14 . by figure 8
it can be shown that , if @xmath254 and @xmath274 are two distinct points in @xmath392 , then the whole segment @xmath399 $ ] belongs to @xmath400 and thus all the points of @xmath392 are colinear .
assume that @xmath401 , @xmath402 , @xmath139 , @xmath403 are the points of @xmath392 that successively on a line @xmath404 , as shown in figure 9 .
since @xmath405 , we have @xmath406 let @xmath407 denote the affine lattice generated by @xmath392 in @xmath404 . in other words , @xmath408 then , @xmath409 is a tiling of @xmath92 .
therefore , as illustrated by figure 9 , there is a @xmath397 such that @xmath410 and consequently @xmath411 * theorem 4 . *
_ let @xmath3 be an @xmath4-dimensional centrally symmetric convex body , let @xmath347 be an @xmath4-dimensional lattice , and let @xmath357 denote the smallest positive number such that @xmath358 is a covering of @xmath2 . if @xmath230 has no segment in the directions of @xmath413 , then the region @xmath349 is a centrally symmetric star body and @xmath367 is a tiling .
_ this result can be proved just like the euclidean case .
let @xmath414 denote the density of the densest lattice packing of @xmath3 and let @xmath1 denote the density of the thinnest lattice covering of @xmath2 by @xmath3 .
theorem 4 has the following corollary .
* remark 6 .
* _ when @xmath418 , there are only five types of minkowski cells @xmath419parallelohedra@xmath420
. therefore , one can determine the values of @xmath421 and @xmath422 by studying a unit ball inscribed in parallelohedra or parallelohedra inscribed in a unit ball .
for other particular nontrivial centrally symmetric convex bodies , for example the octahedron , to enumerate their minkowski cells seems challenging and interesting .
_ it was proved by d.g .
ewald , d.g .
larman and c.a .
rogers @xcite that the line segment directions on the surface of an @xmath4-dimensional convex body is a very small subset of @xmath314 . therefore , theorem @xmath150 and corollary 1 seem can be improved further .
we end this article by three open problems as following .
* problem 1 . *
_ let @xmath159 be an @xmath4-dimensional centrally symmetric polytope .
enumerate the different types @xmath419geometric or combinatorial@xmath420 of the minkowski cells @xmath416 for all lattices @xmath347 . _
* problem 2 . *
_ let @xmath3 be an @xmath4-dimensional centrally symmetric convex body such that all its minkowski bisectors are continuous .
is @xmath423 always a tiling of @xmath2 for all lattice @xmath347 ? _ * acknowledgements . *
this work is supported by 973 programs 2013cb834201 and 2011cb302401 , the national science foundation of china ( no.11071003 ) , and the chang jiang scholars program of china .
for some useful comments , i am grateful to prof .
senlin wu .
bambah , on lattice coverings by spheres , _ proc .
india _ * 20 * ( 1954 ) , 2552 .
barnes , the covering of space by spheres , _ canad .
_ * 8 * ( 1956 ) , 293304 .
u. betke and m. henk , desest lattice packings of @xmath86-polytopes , _ comput .
* 16 * ( 2000 ) , 157186 .
j.h . conway and n.j.a .
sloane , _ sphere packings , lattices and groups _ , ( 3rd edition ) , springer - verlag , new york 1998 .
day , some characterization of inner - product spaces , _ trans .
* 62 * ( 1947 ) , 320337 .
delone and s.s .
ry@xmath427kov , solution of the problem on the least dense lattice covering of a @xmath150-dimensional space by equal spheres , _ dokl .
nauk sssr _ * 152 * ( 1963 ) , 523524 .
m. deza and m. dutour sikiric , voronoi polytopes for polyhedral norms on lattices , arxiv.org : 14010040 .
p. erds , p.m. gruber and j. hammer , _
lattice points _ , longman , essex , 1989 .
ewald , d.g .
larman and c.a .
rogers , the directions of the line segments and of the @xmath428-dimensional balls on the boundary of a convex body in euclidean space , _ mathematika _ * 17 * ( 1970 ) , 1 - 20 .
l. fejes tth , _ lagerungen in der ebene , auf der kugel und in raum _ , springer - verlag , berlin 1972 .
g. fejes tth and w. kuperberg , packing and covering with convex sets , _ handbook of convex geometry _
( p.m. gruber and j.m .
wills , eds . ) , north - holland , amsterdam 1993 , 799860 .
l. few , covering space by spheres , _ mathematika _ * 3 * ( 1956 ) , 136139 .
p.m. gruber , kennzeichnende eigenschaften von euklidischen rumen
und ellipsoiden .
ii , _ j. reine angew . math .
_ * 270 * ( 1974 ) , 123142 .
p.m. gruber and c.g .
lekkerkerker , _ geometry of numbers _
, north - holland , amersterdam 1987 .
d. hilbert , mathematische probleme , _ arch .
phys . _ * 3 * ( 1901 ) , 44 - 63 ; _ bull .
* 37 * ( 2000 ) , 407 - 436 .
horvth , on bisectors in minkowski normaed spaces , _ acta math .
* 89 * ( 2000 ) , 233246 .
hoylman , the densest lattice packing of tetrahedra , _ bull .
* 76 * ( 1970 ) , 135137 .
james , orthogonality in normed linear spaces , _ duke math . j. _ * 12 * ( 1945 ) , 291302 .
lagarias and c. zong , mysteries in packing regular tetrahedra , _ notices amer .
* 59 * ( 2012 ) , 1540 - 1549 . h. mann , untersuchungen ber wabenzellen bei allgemeiner minkowskischer metrik , _ monatsh .
phys . _ * 42 * ( 1935 ) , 417424
. h. martini and k. j. swanepoel , the geometry of minkowski spaces - a survey ( ii ) , _ expos .
_ * 22 * ( 2004 ) , 93 - 144 .
h. minkowski , dichteste gitterfrmige lagerung kongruenter krper , _ nachr .
gttingen , math .- phys .
kl _ ( 1904 ) , 311355 .
rogers , _ packing and covering _ , cambridge university press , cambridge 1964 .
ry@xmath427kov and e.p .
baranovskii , c - type of @xmath4-dimensional lattices and @xmath429-dimensional primitive parallelohedra , _ proc .
steklov inst .
_ * 137 * ( 1978 ) , 1140 .
schmidt , zur lagerung kongruenter krper i m raum , _ monatsh .
_ * 65 * ( 1961 ) , 154 - 158 .
thompson , _ minkowski geometry _ , cambridge university press , cambridge 1996 .
voronoi , nouvelles applications des param@xmath430tres continus @xmath431 la th@xmath432orie des formes quadratiques , deuxi@xmath430me m@xmath432moire , recherches sur les parall@xmath432lo@xmath430dres primitifs , _ j. reine angew . math . _ * 134 * ( 1908 ) , 198 - 287 ; * 135 * ( 1909 ) , 67 - 181 .
woods , a characteristic property of ellipsoids , _ duke math .
j. _ * 36 * ( 1969 ) , 16 . c. zong , _ strange phenomena in convex and discrete geometry _
, springer - verlag , new york 1996 . c. zong , _ sphere packings _ , springer - verlag , new york 1999 . | as a counterpart of hilbert s 18th problem , it is natural to raise the question to determine the density of the thinnest covering of @xmath0 by identical copies of a given geometric object , such as a unit ball , a regular tetrahedron or a regular octahedron . however , systematic study on this problem and its generalizations was started much later comparing with that on packings and the known knowledge on coverings is very limited . by studying minkowski bisectors and minkowski cells , this paper introduces a new way to study the density @xmath1 of the thinnest lattice covering of @xmath2 by a centrally symmetric convex body @xmath3 . some basic results and unexpected phenomena such as example 1 ,
theorems 2 and 4 , and corollary 1 about minkowski bisectors , minkowski cells and covering densities are discovered .
three basic problems about minkowski cells and parallelohedra are presented . |
reaction - diffusion ( rd ) systems are inherent in many branches of physics , chemistry , biology , ecology etc .
the review of the theory and applications of reaction - diffusion systems one can find in books and numerous articles ( see , for example@xcite ) .
the popularity of the rd system is driven by the underlying richness of the nonlinear phenomena , which include stationary and spatio - temporary dissipative pattern formation , oscillations , different types of chemical waves , excitability , bistability etc .
the mechanism of the formation of such type of nonlinear phenomena and the conditions of their emergence have been extensively studied during the last couple decades .
although the mathematical theory of such type of phenomena has not been developed yet due to the essential nonlinearity of these systems , from the viewpoint of the applied and experimental mathematics , the pattern of possible phenomena in rd system is more or less understandable . in the recent years
, there has been a great deal of interest in fractional reaction - diffusion ( frd ) systems @xcite which from one side exhibit self - organization phenomena and from the other side introduce a new parameter to these systems , which is a fractional derivative index , and it gives a greater degree of freedom for diversity of self - organization phenomena . at the same time , the process of analyzing such frd systems is much complicated from the analytical and numerical point of view . in this article , we consider two coupled reaction - diffusion systems : the first one is the classical system @xmath0 @xmath1 where @xmath2 @xmath3 ; @xmath4 , @xmath5 @xmath3 two variables , @xmath6 , @xmath7 @xmath3 are the nonlinear sources of the system modeling their production rates , @xmath8 @xmath3 characteristic times and lengths of the system , @xmath9 @xmath3 is an external parameter and the other model is the fractional rd system @xmath10 @xmath11 with the same parameters and fractional derivatives @xmath12 on the left hand side of equations ( [ 3]),([4 ] ) instead of standard time derivatives , which are the caputo fractional derivatives in time of the order @xmath13 and are represented as @xcite @xmath14 the article is devoted to the second problem and the first one we need for comparing the obtained results with classical one
. equations ( [ 3]),([4 ] ) at @xmath15 correspond to standard rd system described by equations ( [ 1]),([2 ] ) . at @xmath16
they describe anomalous sub - diffusion and at @xmath17 - anomalous superdiffusion . in this paper
, we always assume that the following conditions are fulfilled on the boundaries 0;@xmath18 : \(i ) neumann : @xmath19 \(ii ) periodic : @xmath20 where @xmath21
stability of the steady - state constant solutions of the system ( [ 3]),([4 ] ) correspond to homogeneous equilibrium states @xmath22 can be analyzed by linearization of the system nearby this solution . in this case
the system ( [ 3])([4 ] ) can be transformed to linear system @xmath23 where @xmath24 @xmath25 @xmath26 @xmath27 @xmath28 @xmath29 ( all derivatives are taken at homogeneous equilibrium states ( condition ( [ e ] ) ) . by substituting the solution @xmath30 into frd system ( [ 3]),([4 ] )
we can get the system of linear ordinary differential equations ( [ lin ] ) with the matrix @xmath31 determined by the operator @xmath32 , the stability conditions of which are given by eigenvalues of this matrix .
let us analyze the stability of the solution ( [ e ] ) of the linear system with integer derivatives and find the conditions of this instability ( see , for example:@xcite ) .
we repeat this process in order to compare the results obtained with the results of the fractional rd system considered in this article . in this case , by searching for the solution of the linear system in the form @xmath33 , we get a homogeneous system of linear algebraic equations for constants @xmath34 .
the solubility of this system leads to the characteristic equation @xmath35 where @xmath36 @xmath37 is the identity matrix . as a result ,
the characteristic equation takes on a form of a simple quadratic equation @xmath38 the linear boundary value problem for rd system ( [ 1]),([2 ] ) is unstable according to inhomogeneous wave vectors @xmath39 if ( @xmath40 ) @xmath41 , \label{tu}\ ] ] @xmath42 ( turing bifurcation ) and according to homogenous ( @xmath43 ) fluctuations ( hopf bifurcation ) if ( @xmath44 ) @xmath45 for analyzing the equations ( [ 3]),([4 ] ) let us also consider a linear system obtained near the homogeneous state ( [ e ] ) . as a result
, the simple linear transformation can convert this linear system to a diagonal form @xmath46 where @xmath47 is a diagonal matrix of @xmath31 : @xmath48 eigenvalues @xmath49 are determined by the same characteristic equation ( [ aa ] ) with matrix ( [ a ] ) , @xmath50 @xmath51 is the matrix of eigenvectors of matrix @xmath31 . [
cols="^,^ " , ] the systems have rich dynamics , including steady state dissipative structures , homogeneous and nonhomogeneous oscillations , and spatiotemporal patterns . in this paper
, we focus mainly on the study of general properties of the solutions depending on the value of @xmath52 . as discussed in section 2 , there are two different regions in parameter @xmath53 , where the system can be stable or unstable . in the case of @xmath15 the steady state solutions in the form of nonhomogeneous dissipative structures are inherent to unstable region @xmath54 . figures [ rys6](a)-(c ) show the steady state dissipative structure formation and figures [ rys6 ] ( d)-(f ) present the spatio - temporal evolution of dissipative structures , which eventually leads to homogeneous oscillations . on the figure [ rys6](a)-(d ) , the value @xmath55 increases from @xmath56 to @xmath57 and on this whole interval the structures are in steady state .
this is due to the case @xmath58 , the oscillatory perturbations are damping , and we can see that small oscillations are at the transition period @xmath59 with increasing @xmath55 , the steady state structures change to the spatio - temporary behavior ( figure [ rys6](e)-(f ) ) .
the emergence of homogeneous oscillations , which destroy pattern formation ( figure [ rys6](e),(f ) ) has deep physical meaning .
the matter is that the stationary dissipative structures consist of smooth and sharp regions of variable @xmath60 , and the smooth shape of @xmath61 .
the linear system analysis shows that the homogeneous distribution of the variables is unstable according to oscillatory perturbations inside the wide interval of @xmath62 which is much wider then interval ( @xmath63 ) . at the same time , smooth distributions at the maximum and minimum values of @xmath60 are @xmath64 correspondingly . in the first approximation
, these smooth regions of the dissipative structures resemble homogeneous ones and are located inside the instability regions . as a result ,
the unstable fluctuations lead to homogeneous oscillations , and the dissipative structures destroy themselves .
we can conclude that oscillatory modes in such type fodes have a much wider attraction region than the corresponding region of the dissipative structures .
for a wide range of the parameters @xmath65 , the numerical solutions of the brusselator problem show similar behavior ( figure [ rys7](a)-(d ) ) .
the stationary solutions emerge practically in the same way . at small @xmath65
, we see aperiodic formation of the structures , and approaching @xmath66 , the damping oscillations of the dissipative structures arise . at @xmath67
certain non - stationary structures arise ( figure [ rys7](e ) ) . in this case
, the dissipative structures look quite similar to those we obtained for regular system @xcite .
the increase of @xmath55 leads to a larger amplitude of pulsation .
all these patterns are robust with respect to small initial perturbations .
the further increase of @xmath55 leads to spatially temporary chaos ( figure [ rys7](f ) ) .
in the contrast to previous case , such nonhomogeneous behavior is stable and does not lead to homogeneous oscillations .
the matter is that in brusselator model , the dissipative structures are much grater in amplitude and do not have smooth distribution at the top .
it should be noted that the pulsation phenomena of the dissipative structures is closely related to the oscillation solutions of the ode ( figures [ rys4 ] , [ rys5 ] ) .
moreover , the fractional derivative of the first variable has the most impact on the oscillations emergence .
it can be obtained by performing a simulation where the first variable is a fractional derivative and the second one is an integer .
it should be emphasized that the distribution of @xmath61 , within the solution , only shows a small deviation from the stationary value ( that is why this variable is not represented in the figures ) .
in this article we developed a linear theory of instability of reaction diffusion system with fractional derivatives .
the introduced new parameter
marginal value @xmath66 plays the role of bifurcation parameter .
if the fractional derivative index @xmath52 is smaller than @xmath66 , the system of fodes is stable and has oscillatory damping solutions . at @xmath68 ,
the fodes becomes unstable , and we obtain oscillatory or even more complex - quasi chaotic solutions . in addition , the stable and unstable domains of the system were investigated . by the computer simulation of the fractional reaction - diffusion systems we provided evidence that pattern formation in the fractional case , at @xmath55 less than a certain value , is practicably the same as in the regular case scenario @xmath15 . at @xmath69 ,
the kinetics of formation becomes oscillatory . at @xmath70 ,
the oscillatory mode arises and can lead to homogeneous or nonhomogeneous oscillations . in the last case scenario ,
depending on the parameters of the medium , we can see a rich variety of spatiotemporal behavior .
m. o. vlad and j. ross .
_ systematic derivation of reaction - diffusion equations with distributed delays and relations to fractional reaction - diffusion equations and hyperbolic transport equations : application to the theory of neolithic transition_. phys .
e 66 , 061908 ( 2002 ) .
v.v.gafiychuk , b.y.datsko , yu.yu.izmajlova .
analysis of the dissipative structures in reaction - diffusion systems with fractional derivatives .
metody ta phys .- mech .
v.49 , # 4 , 2006 , pp .
109 - 116 ( in ukrainian ) k.d.oldham and j.spanier _ the fractional calculus : theory and applications of differentiation and integration to arbitrary order , vol.111 of mathematics in science and engineering _ academic press , new york , 1974 .
v.v.gafiychuk , b.s.kerner , v.v.osipov , t.m.scherbatchenko .
formation of pulsating thermal - diffusion autosolitons and turbulence in a nonequilibrium electron - hole plasma .
phys . sem .
v.25 , # 11 , ( 1991 ) ( translation of fiz.tekh .
poluprovodn ( ussr ) .
v.25 , no.11 , ( 1991 ) , pp .
16961702 ) .
v.v.gafiychuk , a.v.demchuk .
analysis of the dissipative structures in gierer - meinhardt model .
matematicheskie metody i physico - mechanicheskie polia .
v.40 , # 2 , 1997 , pp.48 - 53 ( in ukrainian , english translation in journal of mathematical sciences , v.88 , # 4 , 1998 ) . | this paper is concerned with analysis of coupled fractional reaction - diffusion equations .
it provides analytical comparison for the fractional and regular reaction - diffusion systems . as an example , the reaction - diffusion model with cubic nonlinearity and brusselator model are considered .
the detailed linear stability analysis of the system with cubic nonlinearity is provided .
it is shown that by combining the fractional derivatives index with the ratio of characteristic times , it is possible to find the marginal value of the index where the oscillatory instability arises .
computer simulation and analytical methods are used to analyze possible solutions for a linearized system .
a computer simulation of the corresponding nonlinear fractional ordinary differential equations is presented .
it is shown that the increase of the fractional derivative index leads to the periodic solutions which become stochastic at the index approaching the value of 2 .
it is established by computer simulation that there exists a set of stable spatio - temporal structures of the one - dimensional system under the neumann and periodic boundary condition .
the characteristic features of these solutions consist in the transformation of the steady state dissipative structures to homogeneous oscillations or space temporary structures at a certain value of fractional index
. reaction - diffusion system , fractional differential equations , oscillations , dissipative structures , pattern formation , spatio - temporal structures 37n30 , 65p40 , 37n25 , 35k50 , 35k45 , 34a34 , 34c28 , 65p30 |
the study of temporal chaos in low - dimensional systems , some of which can be described by low - dimensional maps @xcite , was extremely beneficial for the understanding of turbulence . in 1984
coupled map lattices were introduced into the physical literature as a tool for studying spatiotemporal chaos in spatially extended , i.e. high - dimensional systems @xcite .
they consist of spatially coupled low - dimensional maps and represent dynamical systems that are discrete in space and time , but continuous in the state variable .
they serve as models for coupled josephson junctions , excitable media , population dynamics , neural dynamics and turbulence @xcite .
although bunimovich and sinai mathematically proved a number of statements regarding the appearance of coherent structures from spatiotemporal chaos @xcite , most results in the field have been obtained by numerical simulations @xcite . while the study of temporal chaos has greatly profited from the existence of simple maps like the bernoulli shift map and the cat map @xcite , which can be solved explicitly ( for integer expansion rates ) , thereby making the mechanisms of mixing and temporal chaos understandable ,
no investigation of this type has been provided up to now for the problem of spatiotemporal chaos . here
we present a solvable model for spatiotemporal chaos which can be solved in the sense of the bernoulli map .
we start from the model equations @xcite @xmath0,\ ] ] where the index @xmath1 runs over the @xmath2 sites of a discrete lattice , @xmath3 measures the strength of the spatial coupling between neighboring sites and @xmath4 is a local map which determines the evolution of the continuous variable @xmath5 at discrete time - steps @xmath6 .
the extension of eq .
( [ eq_cml_gen ] ) to more than nearest neighbor coupling and arbitrary dimensions is straightforward .
we first write it in the new variables @xmath7 as @xmath8.\ ] ] in order to obtain a solvable model we choose @xmath9 , i.e. the bernoulli shift map with integer stretching factor @xmath10 . to make our approach for the spatially extended system more transparent we will calculate in section [ sec_single ] the invariant density and a temporal correlation function for the single bernoulli shift map . in section [ sec_lattice ]
we will show that the coupled map lattice is solvable for special values of the coefficients @xmath10 and @xmath3 in eq .
( [ eq_cml ] ) and we will also calculate the invariant density and a spatial and a spatiotemporal correlation functions .
finally in section [ sec_discuss ] we discuss our results and indicate directions of further research .
first we recall that the single map @xmath11 can be solved as @xmath12 because @xmath13 \right\ } { \rm mod}\;1\cr = & & \left\ { a\left [ ax^{0}-k^{0}\right ] \right\ } \,{\rm mod}\;1=(a^{2}x^{0})\,{\rm mod}\;1 \end{aligned}\ ] ] where @xmath14 is an integer which represents the action of the modulo function . for the last equality sign in to hold we needed the fact that @xmath10 is an integer such that @xmath15 becomes again an integer , which can be dropped within the last modulo function .
since the modulo function confines the variable @xmath16 to a circle we could view the bernoulli shift map as a linear map @xmath17 where the variables live on a unit circle , i.e. on a 1-torus .
we shall see below that we can view our coupled map system as a linear map acting on variables confined to an @xmath2-torus , where @xmath2 is the number of lattice sites .
the invariant density @xmath18 of the simple bernoulli shift map measures the distribution of @xmath19 values on the attractor generated by the map and is well known to be a constant @xcite .
we can obtain this result by noting that @xmath20 is defined on an unit circle , i.e. it is periodic in @xmath19 and therefore can be represented as a fourier series @xmath21 where @xmath22 takes only integer values @xmath23 .
the invariant density @xmath18 evolves from an initial distribution @xmath24 according to the frobenius - perron equation @xcite @xmath25 \rho^0\left ( x^{\prime } \right)\ ] ] and is defined as the long - time limit @xmath26 . in order to solve eq .
( [ eq_dens_sing ] ) we use ( [ eq_four_sing ] ) and the fact that the bernoulli shift map becomes a linear map on a torus , i.e. @xmath27)=\exp(2\pi ia^{t}x)$ ] , and obtain @xmath28 if we make the reasonable assumption that the initial distribution @xmath24 is non - singular , then @xmath29 .
this means that in eq .
( [ eq_four_sing ] ) all fourier coefficients of @xmath30 tend to zero in the infinite - time limit , except the one , which belongs to @xmath31 . since @xmath32 this yields @xmath33 and @xmath34 . in a similar fashion
we can now define and calculate the time correlation function on the 1-torus .
the usual time correlation function is defined via @xmath35 where the time evolution of @xmath19 is given by the map @xmath36 , but here we introduce the time correlation function @xmath37 @xmath38 where @xmath39 which respects the fact that the variable @xmath19 is an angular variable on a torus @xcite . for @xmath34
we obtain from eqs .
( [ eq_sing_cor],[eq_sing_sol ] ) @xmath40 in the following section we will demonstrate what changes have to be made in order to compute in a similar fashion as above the solution of the dynamical equations , the invariant density and the time correlation function for our coupled map lattice .
for the bernoulli shift map @xmath9 the time evolution for the variables @xmath41 of the coupled map lattice becomes according to eq .
( [ eq_cml ] ) @xmath42\right)\,{\rm mod}\;1.\ ] ] if the parameters @xmath43 and @xmath44 are such that both @xmath45 and @xmath46 take integer values @xmath47 and @xmath48 , then the equation of motion for the coupled map system can be written in the compact form @xmath49 where the coupling matrix @xmath50 has integer elements @xmath51 we will now free ourselves from the specific form ( [ eq_mat_coup ] ) for @xmath52 , which was physically motivated by the nearest neighbor lattice model ( [ eq_map_nnc ] ) and show that eq .
( [ eq_map_mat ] ) can be solved for _ any _ matrix @xmath50 , which has _ integer valued elements _ @xmath52 . in order to see this we write ( [ eq_map_mat ] ) in vector notation as @xmath53 where @xmath54 and the modulo is taken for each component of the vector @xmath55
. then we obtain by iterating from the initial condition : @xmath56 where @xmath57 is a vector with integer components which represents the action of the modulo function .
this yields @xmath58 \right){\rm mod}\;1 \cr = & & \left ( aa{\bf x}^t - a{\bf k}^t\right){\rm mod}\;1 = \left ( a^2{\bf x}^t\right){\rm mod}\;1 , \end{aligned}\ ] ] where the last equality sign only holds because all elements of the matrix @xmath50 are integers , such that @xmath59 is a vector with integer components which can be dropped under the last modulo function .
since ( [ eq_mod_iter ] ) holds for any @xmath60 , we obtain the closed - form solution as a function of the initial value @xmath61 eq . ( [ eq_mod_iter ] ) shows that we can solve not only our coupled map lattice problem ( [ eq_map_nnc ] ) , but all linearly coupled systems , where the coupling occurs via a matrix @xmath50 with integer elements @xmath52 and the nonlinearity is provided by the modulo function .
the solution can be obtained by first solving the linear problem , i.e. by obtaining @xmath62 and then taking the modulo , which is the same as having the linear map acting on an @xmath2-torus in analogy to the famous arnold s cat map in two dimensions @xcite .
next we investigate the invariant density and the spatiotemporal correlation function of the coupled map lattice .
the first quantity gives us information about the measurable time averaged spatial structures in the system and the second one tells us about the measurable spatiotemporal structures .
the invariant density @xmath63 yields the distribution of points on the attractor generated by the map @xmath64 . by starting from an initial distribution
@xmath65 it could be obtained as the infinite - time limit of @xmath66 in the frobenius - perron equation @xmath67 \rho^0({\bf x}')\ ] ] since all quantities involved in eq .
( [ eq_dens_ev ] ) are periodic on an @xmath2-torus , the fourier decomposition of @xmath66 contains only wavevectors @xmath68 with integer components , i.e. @xmath69 by using the equality @xmath70 ) = \exp(2\pi i[(a^t)^t{\bf k}]\cdot{\bf x})$ ] eq .
( [ eq_dens_four ] ) yields @xmath71 if the initial distribution @xmath65 is non - singular , all fourier coefficients vanish for large values of the wavevector , i.e. @xmath72 for a completely expanding map , where all eigenvalues of the matrix @xmath50 have an absolute value larger than one , @xmath73 for each @xmath74 and the only non - vanishing fourier component becomes @xmath75 which yields a constant invariant density @xmath76 this result is completely analogous to the single map case .
however we may obtain different results for the invariant density if there are contracting directions in the phase space .
indexing the stable and unstable eigenvalues @xmath77 and right ( left ) eigenvectors @xmath78 ( @xmath79 ) of the matrix @xmath80 with indices `` s '' and `` u '' respectively , we have @xmath81 according to the above , we will only obtain results which differ from the trivial expanding case , if there exists at least one @xmath74 , such that its components along the unstable directions are all zero , i.e. it is contained in the stable manifold @xmath82 of the fixed point @xmath83 of the `` conjugate '' map @xmath84 in fact the above argument does not take into account one particular feature of the system .
specifically , the map ( [ eq_conj_map ] ) might not be hyperbolic , i.e. it might possess the _
manifold , defined by the eigenvectors corresponding to @xmath85 .
if this is the case ( and it is for the coupled map lattice ( [ eq_map_nnc ] ) as we shall see below ) , we assume that the invariant density , that we calculate corresponds to the _ physical _ ( or kolmogorov ) _ measure _ @xcite .
the latter is calculated by introducing a small amount of noise into the system and then taking the zero - noise limit .
assuming that the physical measure is unique , we can obtain the invariant density starting from @xmath86 , which is equivalent to @xmath87 .
then only the stable manifold contributes to the invariant density and the central manifold can be treated as the unstable one .
this essentially means , that the averaging over the central directions is accomplished by the infinitely small noise present in the system . on the other hand , eqs .
( [ eq_fou_comp],[eq_eig_exp ] ) tell us that it is not enough to have contracting eigenvalues in order to get a non - constant invariant density .
let us first consider the case with a single stable direction @xmath88 . since all components @xmath89 of a vector @xmath68 are integer , it is contained within the stable manifold @xmath82 only if @xmath90 .
this in turn means that the components @xmath91 should be mutually rational , i.e. @xmath92 an example where we have one contracting and one expanding direction is the cat map @xmath93 although the eigenvalue corresponding to the contracting direction is @xmath94 , this map still has a constant invariant density because the components of the eigenvector @xmath95 belonging to @xmath96 have a non - rational ratio ( see fig .
[ fig_cat ] ) leading to @xmath97 generally , in order to get a non - constant invariant density our model must possess a stable manifold , which in turn should contain at least one vector with mutually rational components .
every vector @xmath98 with integer components , which is pulled in the long time limit into the origin , according to eq .
( [ eq_eig_exp ] ) , @xmath99 can be represented as a linear combination of a ( usually small ) number of basis vectors @xmath100 , @xmath101 with integer coefficients @xmath102 , i.e. @xmath103 .
the invariant density will only contain non - vanishing fourier components with such wavevectors @xmath104 , all with weight 1 : @xmath105 up to now our conclusions have been completely general for any coupling matrix @xmath50 with integer elements .
let us now consider the condition ( [ eq_pull_in ] ) in more detail for our 1-dimensional nearest neighbor model ( [ eq_map_nnc ] ) .
the corresponding matrix ( [ eq_mat_coup ] ) can be diagonalized by fourier transformation in the space variables @xmath1 , leading for periodic boundary conditions to eigenvalues @xmath106 and the corresponding eigenvectors @xmath107 where @xmath108 and @xmath109 . of these
only a few have mutually rational components .
for instance , both @xmath110 and @xmath111 are rational only if @xmath112 is rational , which immediately restricts the allowed wavevectors @xmath108 to a set of 5 values : @xmath113 .
each @xmath114 generates several basis vectors , provided @xmath115 : @xmath116 rationality of @xmath117 in not an unexpected result , e.g. choosing @xmath118 results in the eigenvalue @xmath119 , according to ( [ eq_evalue ] ) , which requires @xmath120 where we defined @xmath121 with @xmath122 if @xmath123 and @xmath124 otherwise .
this in turn , requires @xmath125 , which is seen to be the case by rewriting ( [ eq_inv_dens ] ) as @xmath126 it is useful to define the projection of the invariant density @xmath127 on a chosen direction @xmath128 : @xmath129 for example , if @xmath130 , eq . ( [ eq_project ] ) gives the distribution of the @xmath131-th map variable @xmath132 .
if @xmath128 coincides with one of the basis directions , i.e. @xmath133 for some @xmath134 , the projection @xmath135 ( @xmath136 denotes the unit @xmath2-dimensional cube ) becomes singular : we get a series of @xmath137-functions with an envelope @xmath138 otherwise , the projection ( [ eq_project ] ) is a continuous , non - singular function of parameter @xmath139 .
in other words , only the projection on the directions defined by the basis vectors @xmath100 is singular .
in particular , the eigenvector @xmath140 defines a basis direction @xmath141 if and only if the projection ( [ eq_project ] ) on this eigenvector ( we define @xmath142 for @xmath143 ) , @xmath144 is singular .
this implies that @xmath145 for some @xmath131 .
one can trivially verify that the projection ( [ eq_eig_proj ] ) has the average @xmath146 and the dispersion given by @xmath147 for all @xmath148 and almost always for @xmath149 .
a few degenerate cases like @xmath150 or @xmath151 for @xmath152 , or @xmath153 for @xmath154 give different dispersions .
as expected , numerically calculating the projection @xmath155 on the stable and unstable directions ( [ eq_evector ] ) , we only get a singular distribution for @xmath156 ( fig . [ fig_narrow](b ) ) , provided that the respective eigenvector is stable ( @xmath157 ) .
otherwise a smooth gaussian - like distribution is obtained ( fig .
[ fig_narrow](a ) ) .
indeed one can easily see that in the large - length limit both the continuous distribution and the envelope of the singular distribution ( [ eq_eig_proj ] ) become gaussian : @xmath158 where @xmath159 is the normalized gaussian and @xmath160 .
the standard spatial correlation function is trivially calculated to yield @xmath161 for the completely expanding case with @xmath162 ( here @xmath163 denotes the average taken with @xmath63 ) . if there are contracting directions , we rewrite ( [ eq_corr_fun ] ) as @xmath164 where @xmath165 for all @xmath166 except @xmath167 . since @xmath168 for all @xmath148 , @xmath169 this reduces to a @xmath137-correlation ( which coincides with the result ( [ eq_corr_fun ] ) obtained for @xmath162 ) in all but a few special cases , when @xmath170 .
for instance , choosing @xmath171 and @xmath172 yields for @xmath154 @xmath173 and therefore @xmath174 and @xmath175 , resulting in @xmath176 since the invariant density , although being nontrivial , does not tell us much about the spatiotemporal structures in the system , next we introduce a spatiotemporal correlation function @xmath177 , which is a straightforward generalization of the time correlation function ( [ eq_sing_cor ] ) : @xmath178 by expanding @xmath63 into fourier series we obtain in analogy to ( [ eq_fou_comp ] ) : @xmath179 since only the non - vanishing fourier components @xmath180 ( where @xmath181 ) of the invariant density ( [ eq_inv_dens ] ) contribute , ( [ eq_time_four ] ) reduces to @xmath182 in a translationally invariant system @xmath177 does not depend on @xmath1 , so we drop the index and fix @xmath1 ( set @xmath183 to be specific ) .
it can be easily verified that the correlation ( [ eq_time_cor ] ) is short - ranged in both space and time .
first we note that it vanishes if the vector @xmath184 with components @xmath185 does not lie on the stable manifold @xmath82 .
according to ( [ eq_eig_exp ] ) @xmath186 where @xmath187 is the largest and @xmath188 the next largest eigenvalue and @xmath189 and @xmath190 are the respective eigenvectors . for increasing @xmath60
the vector @xmath184 asymptotically approaches the direction defined by @xmath189 and therefore can not lie on the stable manifold for @xmath191 , where @xmath192 is some finite ( and typically small ) integer .
on the other hand for @xmath193 we have @xmath194 since all basis vectors ( [ eq_basis ] ) are periodic in space with periods 1 , 2 , 3 , 4 or 6 , any linear combination of these will also be periodic with period of at most 12 .
since the vector with components @xmath195 is _ not _ periodic for @xmath196 , the maximal size of the system with non - trivial correlation is limited by @xmath197 .
again , choosing @xmath171 and @xmath172 for @xmath154 as an example , we have @xmath198 and @xmath199 as basis vectors and consequently @xmath200 i.e. we retrieve the result ( [ eq_corr_4 ] ) .
to summarize , we have shown that the solution for the dynamical behavior of a lattice of bernoulli maps that are coupled by a matrix @xmath50 with integer coefficients can be given in the closed form as @xmath201 : the dynamical behavior of the coupled map system can be described by the repeated action of a linear map @xmath62 on variables that are confined to an @xmath2-torus .
this picture explains that the essentials of the dynamical behavior are dictated by the eigenvalues and eigenvectors of @xmath50 .
the invariant density @xmath63 of this system displays fourier coefficients that are different from zero , i.e. is non - constant , whenever the stable manifold of the zero wavevector contains a non - empty basis of directions @xmath100 with mutually rational components , generating the infinite asymptotically contracting set of wavevectors . for nearest neighbor couplings in a 1-dimensional lattice ( given by eq .
[ eq_mat_coup ] ) the maximal number of basis vectors is eight ( actually even less , since , e.g. @xmath202 ) . we have calculated the standard spatial correlation function @xmath203 for the model with nearest neighbor couplings and shown that it is given by @xmath204 almost always .
a few special cases exist however for sufficiently small lattices , where the spatial correlations are different .
nevertheless , @xmath203 always vanishes at sufficiently large distances . the invariant density of this system and the spatial correlation function display little structure as compared to the lyapunov spectrum , which is , for the nearest neighbor coupling , given by @xmath205 .
this result shows that the time averaged spatial behavior is _ not _ simply a straightforward reflection of the lyapunov spectrum ( see related work listed in @xcite ) .
it is instructive to compare our results with the general results obtained by bunimovich and sinai @xcite , who proved that , for sufficiently small coupling ( in our case determined by parameter @xmath3 ) , certain expanding coupled map systems with finite - range coupling possess an absolutely continuous invariant measure @xmath207 , and also that the time and space correlation functions decay exponentially ( not slower than exponentially , to be exact ) .
our results indicate , that for larger coupling , the invariant measure still exists , but might not be absolutely continuous due to the fact , that large coupling often causes the appearance of contracting directions , even if the isolated local maps @xmath36 are expanding .
the space and time correlations in our model are seen to decay even faster then exponentially , but the few special cases , giving non - trivial correlations imply that there might be some general relationship between the continuity of the invariant measure and the appearance of coherent structures in the system .
one open problem is the extension of our results to higher dimensions and to couplings which have a longer range . in the 1-dimensional case
the eigenvectors remain also valid for longer - ranged couplings , only the eigenvalues change .
this means that a model with long , but finite , range will have no more structure in the invariant density than the short - ranged model .
this is of course a peculiarity of the bernoulli shift map , but should again be taken as a warning for making conclusions from the spatial range of the coupling onto the observable spatial patterns .
although our solution for the dynamics and the correlation functions hold for general dimensions it would be interesting to see what the restrictions on the wavevectors that generate the basis of the invariant density look like in two and three dimensions .
finally , one could investigate the dynamical behavior of a system , whose time dependence is given a priori by the equation ( [ eq_solution ] ) also for matrices @xmath50 with _ non - integer _
elements . by doing so
one will loose the property of the original map , that the relation relation ( [ eq_map_vec ] ) holds step by step , but the trajectories generated by equation ( [ eq_solution ] ) are well defined .
thanks c. koch for the kind hospitality extended to him at caltech and the volkswagen foundation for financial support .
the authors thank m. c. cross for the careful reading of the manuscript .
this research has also been partially supported by the nsf through grant no .
dmr-9013984 .
l. a. bunimovich and ya .
g. sinai , nonlinearity * 1 * , 491 ( 1988 ) ; statistical mechanics of coupled map lattices , in _ theory and applications of coupled map lattices _
k. kaneko , j. wiley & sons , singapore , 1993 ) .
h. daido , prog .
* 72 * , 853 ( 1984 ) ; f. kaspar and h. g. schuster , phys . lett .
a * 113 * , 451 ( 1986 ) ; t. bohr and o. b. christensen , phys .
. lett . * 63 * , 2161 ( 1989 ) ; c. nicolis , g. nicolis and q. wang , bifurcations and chaos * 2 * , 263 ( 1992 ) ; w.yang , e. j. ding , m. ding , phys . rev . lett . * 76 * , 1808 ( 1996 ) . | we show that the dynamical behavior of a coupled map lattice where the individual maps are bernoulli shift maps can be solved analytically for integer couplings .
we calculate the invariant density of the system and show that it displays a nontrivial spatial behavior .
we also introduce and calculate a generalized spatiotemporal correlation function . |
exclusive processes with hadronic final states test various model calculations motivated by perturbative or non - perturbative qcd .
two - photon production of exclusive hadronic final states is particularly attractive due to the absence of strong interactions in the initial state and the possibility of calculating @xmath15 amplitudes .
the perturbative qcd calculation by brodsky and lepage ( bl ) @xcite is based on factorization of the amplitude into a hard scattering amplitude for @xmath16 and a single - meson distribution amplitude .
their prediction gives the dependence on the center - of - mass ( c.m.)energy @xmath17 and scattering angle @xmath18 for @xmath19 processes @xmath20 where @xmath21 represents a meson and @xmath22 denotes its electromagnetic form factor .
vogt @xcite , based on the perturbative approach , claimed a need for soft contributions , as his result for the hard contribution was well below the experimental cross section obtained by cleo @xcite .
diehl , kroll and vogt ( dkv ) proposed @xcite the soft handbag contribution to two - photon annihilation into pion or kaon pairs at large energy and momentum transfers , in which the amplitude is expressed by a hard @xmath23 subprocess and a form factor describing the soft transition from @xmath24 to the meson pair .
dkv , as well as bl , predict the @xmath10 dependence of the angular differential cross section , which is an important test of these approaches .
it is interesting to investigate experimentally an energy scale where those theoretical predictions become valid .
the recent measurements of @xmath13 and @xmath1 performed by aleph @xcite are consistent , within their errors , with the bl s prediction of the energy dependence , but not the normalization .
however , their dataset is not sufficient to conclusively test the @xmath4 and @xmath10 dependences . in this report
, we measure @xmath13 and @xmath14 processes with high precision , and make quantitative comparisons with qcd predictions .
this analysis is based on an @xmath2 data sample collected at or near the @xmath25 resonance energy , accumulated with the belle detector @xcite located at kekb @xcite .
kekb is a colliding beam accelerator of 8@xmath26 electrons and 3.5@xmath26 positrons designed to produce copious @xmath27 meson pairs to observe _ cp _ violation . the belle detector , with a @xmath28 solenoidal magnetic field , surrounds the interaction point and subtends the polar angle range @xmath29 , measured from the @xmath30 axis , which is aligned opposite the positron beam .
it is described in detail in ref .
briefly , charged track momenta and their decay points are measured by the central drift chamber ( cdc ) and silicon vertex detector ( svd ) .
the hadron identity of these charged tracks is determined using information from the time - of - flight counters ( tof ) , the aerogel threshold erenkov counters ( acc ) , and the specific ionization in the cdc .
hadron / electron discrimination is performed using the above information as well as the energy deposition and shower profile in the segmented csi(tl)-crystal electromagnetic calorimeter ( ecl ) .
hadron / muon discrimination is achieved using information from the neutral - kaon and muon detector ( klm ) , which consists of glass resistive plate counters embedded in the solenoid s iron flux return .
the signal events are collected predominantly by a trigger that requires two charged tracks penetrating through the cdc and tof , with an opening angle in the @xmath31 plane ( perpendicular to the @xmath30 axis ) of at least @xmath32 .
we select signal candidates according to the following criteria .
there must be exactly two oppositely - charged reconstructed tracks satisfying the following conditions : @xmath33 for the polar angle @xmath34 of each track ; @xmath35 gev/@xmath36 for the momentum component in the @xmath31 plane of each track ; @xmath37 cm and @xmath38 cm for the origin of each track relative to the nominal @xmath39 collision point ; and @xmath40 cm for the two tracks origin difference along the @xmath30 axis , where the origin is defined by the closest approach of the track to the nominal collision point in the @xmath31 plane .
the event is vetoed if it contains any other reconstructed charged track with transverse momentum above 0.1 gev/@xmath36 .
cosmic rays are suppressed by demanding that the opening angle @xmath41 between the two tracks satisfy @xmath42 .
the signal is enriched relative to other backgrounds by requiring that the scalar sum of the momentum of the two tracks be below 6 gev/@xmath36 , the total energy deposited in the ecl be below 6 gev , the magnitude of the net transverse momentum of the above - selected two charged tracks in the @xmath39 c.m .
frame be below 0.2 gev/@xmath36 ( this condition is tightened in the next section ) , the invariant mass of these two tracks be below 4.5 gev/@xmath43 , and that the squared missing mass of the event be above 2 gev@xmath44/@xmath45 . here , the two tracks are assumed to be massless particles .
the latter two requirements eliminate radiative bhabha and initial state radiation events .
the remaining events consist of two - photon production of @xmath39 , @xmath46 , @xmath47 , @xmath48 , and @xmath49 final states as well as unvetoed @xmath50 events according to a monte carlo ( mc ) study @xcite .
the predicted versus measured range and transverse deviation of hits in the klm are used to construct a normalized likelihood @xmath51 that a track extrapolated from the cdc is a muon rather than a pion or kaon . here
, we classify an event as arising from @xmath52 if either track has @xmath53 .
similarly , the tof , acc , cdc , and ecl information is used to construct a normalized likelihood @xmath54 that a reconstructed track is an electron rather than a hadron .
we classify an event as arising from @xmath55 if either track has @xmath56 . for above two separations 93% of signal events
survive for both modes according to the mc study described later .
the tof , acc and cdc information is used to construct another normalized likelihood @xmath57 that a reconstructed track is a proton rather than a kaon , with a high value corresponding to a proton - like track .
the scatterplot of this quantity for the negative track _ vs. _ that for the positive track in each event is shown in fig .
[ fg_sel](a ) .
note that the peak near the origin contains both @xmath1 and @xmath0 candidates .
events above the hyperbolic curve @xmath58 , shown in the inset of fig .
[ fg_sel](a ) , are deemed to arise from @xmath59 .
cc @xmath60 and ( b ) @xmath61@xmath62 .
the cut boundaries are shown in the top - view insets . ]
@xmath60 and ( b ) @xmath61@xmath62 .
the cut boundaries are shown in the top - view insets . ] after removing events that appear to arise from two - photon production of @xmath46 , @xmath39 , and @xmath49 according to the above criteria , the remaining sample consists of two - photon production of @xmath48 , @xmath47 , and residual @xmath46 , as well as @xmath50 production where each @xmath63 lepton decays to a single pion or muon .
information from the tof , acc , and cdc is used to form a normalized likelihood ratio @xmath64 that a reconstructed track is a kaon rather than a pion ( or muon ) , with a high value corresponding to a kaon - like track . the scatter plot of this quantity for the negative track _ vs. _ that for the positive track is shown in fig . [ fg_sel](b ) .
note that the peak near the origin contains @xmath0 , @xmath65 , and residual @xmath46 events .
events above the diagonal line @xmath66 , shown in the inset of fig .
[ fg_sel](b ) , are classified as @xmath1 candidates , while events below the line @xmath67 are classified as @xmath0 candidates ( including @xmath65 and residual @xmath68 backgrounds ) .
events in the diagonal band between these two lines are discarded .
the @xmath0 sample is somewhat contaminated by non - exclusive two - photon background @xmath69 as well as the @xmath50 process , in roughly equal proportion .
we note that these backgrounds appear at high values of the magnitude of the net transverse momentum @xmath70 in the @xmath39 c.m .
frame , and are often accompanied by photons from the prompt decay of a neutral pion in the final state . therefore , we reject events in the @xmath0 sample that contain a photon with energy above 400 mev ( @xmath71-veto ) .
the distributions of @xmath70 for the @xmath0 candidates before and after application of this veto are shown as the histograms in fig .
[ fg : ptb](a ) .
the yields of the @xmath47 and @xmath48 events are expressed as functions of three variables : @xmath4 derived from the invariant mass of the two mesons , @xmath72 and @xmath70 .
eighty - five 20 mev wide bins in @xmath4 times six bins in the cosine of the @xmath73 c.m .
scattering angle @xmath18 times twenty bins in net transverse momentum are used in the ranges @xmath5 , @xmath6 , and @xmath74 .
the spectrum of the residual @xmath75 background within the @xmath0 sample is obtained from a mc simulation , based on a full @xmath76 qed calculation @xcite , with a data sample corresponding to an integrated luminosity of @xmath77 that is processed by the full detector simulation program and then subjected to trigger simulation and the above event selection criteria .
after calibration of the muon identification efficiency to match that in the data using identified @xmath75 events , the residual @xmath68 background is scaled by the integrated luminosity ratio and then subtracted .
the excess in the @xmath78-vetoed histogram of fig .
[ fg : ptb](a ) above the smooth curve from the signal mc described in more detail below is attributed to non - exclusive @xmath79 events that are not rejected by the @xmath71-veto ; most of the @xmath50 events are rejected by this veto .
a similar excess appears in fig .
[ fg : ptb](b ) for the @xmath14 process .
assuming that this remaining background is proportional to net transverse momentum , we determine the slope using the difference between data and mc in the range @xmath80 gev/@xmath36 for each 200 mev wide @xmath4 bin , then smooth the so - determined slopes by fitting them to a cubic polynomial in @xmath4 .
we verify that there is no dependence on the scattering angle @xmath18 . using the smoothed slope ,
the estimated non - exclusive background is subtracted from each bin .
finally , we restrict our signal region to net transverse momentum below 0.05 ( 0.10 ) gev/@xmath36 for @xmath13 ( @xmath14 ) . the background - subtracted yields , integrated over net transverse momentum and scattering angle , are shown as a function of @xmath4 in fig .
[ fg : kkpicrtw ] .
distribution for @xmath0(a ) and @xmath1(b ) candidates .
the dashed and solid histograms in @xmath0 indicate the distribution of events before and after @xmath71-veto ( which is not applied to the @xmath1 candidates ) , respectively .
the arrows indicate the upper boundaries of @xmath81 for the signal .
the residual muon background has been subtracted from the @xmath0 distribution .
the curves show the signal mc distribution which is normalized to the signal candidates at the leftmost bin . ] ) obtained for the @xmath13(solid ) and @xmath14(dashed ) samples after the background subtraction . ]
the signatures of the @xmath82 and @xmath83 resonances are observed in both @xmath0 and @xmath1 channels . by fitting each @xmath4 distribution outside the range 3.33.7 gev to a cubic polynomial
, we see an excess of @xmath84 ( @xmath85 ) events in the @xmath0 ( @xmath1 ) channel in the @xmath11 range of 3.343.44 gev and a corresponding excess of @xmath86 ( @xmath87 ) events in the @xmath12 range of 3.543.58 gev .
we obtain consistent results from a fit of each distribution to a cubic polynomial plus a breit - wigner ( @xmath11 ) or a gaussian ( @xmath12 ) peak . assuming a flat ( @xmath88 ) shape for the @xmath11 ( @xmath12 ) resonance @xcite
, we subtract the above excesses bin by bin from each angular distribution in the above @xmath4 ranges .
the @xmath11 statistical significance is @xmath89 ( @xmath90 ) in the @xmath0 ( @xmath1 ) channel , where @xmath91 is the standard deviation .
the @xmath12 statistical significance is @xmath92 ( @xmath93 ) in the @xmath0 ( @xmath1 ) channel .
the significances are taken from the square root of the difference of the goodness - of - fit values from the two fits where the peak term in the above fit function is included or excluded .
the differential cross section for a two - photon process to a two - body final state arising from an electron - positron collision is given by @xmath94 where @xmath95 and @xmath96 denote the number of the signal events and a product of detection and trigger efficiencies , respectively ; @xmath97 is the integrated luminosity , and @xmath98 is the luminosity function , defined as @xmath99 .
the efficiencies @xmath100 for @xmath13 and @xmath14 are obtained from a full monte carlo simulation @xcite , using the treps @xcite program for the event generation as well as the luminosity function determination .
the trigger efficiency is determined from the trigger simulator .
the typical value of the trigger efficiency is @xmath101 for events in the acceptance .
the efficiency - corrected measured differential cross sections for @xmath13 and @xmath14 , normalized to the partial cross section @xmath102 for @xmath6 , are shown in fig .
[ fg : sin4 ] for each 100 mev wide @xmath4 bin .
the partial cross sections @xmath102 for both processes , integrated over the above scattering angle range , are shown in fig .
[ fg_kkpisigmaw ] ( along with their ratio ) and itemized in table [ table_sigma ] . .
cross sections and errors for the @xmath13 and @xmath14 processes , in the angular range @xmath6 . [ cols="^,^,^,^,^,^,^ " , ] [ tb : gggbr ] the measured yields of @xmath11 and @xmath12 events can be used to extract the two - photon decay width of each resonance using the formula @xmath103 where @xmath104 and @xmath105 are the yield and mass , respectively , of the corresponding spin-@xmath106 @xmath107 resonance ( where the mass is taken from ref .
@xmath96 denotes the product of the detection efficiency and acceptance for the resonance decays .
the extracted values of @xmath108 are summarized in table [ tb : gggbr ] .
the ratios @xmath109 are consistent with the known @xmath110 @xcite ratios for both resonances .
therefore we combine the measurements of the widths from the two modes to obtain @xmath111 and @xmath112 where the error associated with the branching fraction ( taken from @xcite ) is calculated assuming independent uncertainties for the @xmath47 and @xmath48 modes .
these values are consistent with the two - photon decay widths of @xmath11 and @xmath12 following from the total widths and branching ratios into two photons obtained in previous experiments @xcite .
using @xmath2 of data collected with the belle detector at kekb , we have measured with high precision the cross sections for the @xmath13 and @xmath14 processes in the two - photon c.m .
energy range @xmath5 and angular range of @xmath6 .
the angular differential cross sections for those processes show a @xmath10 behavior in the range of @xmath9 , which is predicted by qcd .
the energy dependence of the differential cross section is also consistent with the qcd prediction of @xmath113 .
the partial cross section ratio @xmath114 , in @xmath6 is measured to be @xmath8 for @xmath9 in accordance with the prediction based on the qcd sum rules and su(3 ) symmetry breaking .
we have observed the @xmath11 and @xmath12 resonances decaying to @xmath0 and @xmath1 final states and measured the products of their two - photon decay widths and two - hadron branching fractions for the first time .
* acknowledgements * + we thank the kekb group for the excellent operation of the accelerator , the kek cryogenics group for the efficient operation of the solenoid , and the kek computer group and the national institute of informatics for valuable computing and super - sinet network support .
we are grateful to v. chernyak for fruitful discussions .
we acknowledge support from the ministry of education , culture , sports , science , and technology of japan and the japan society for the promotion of science ; the australian research council and the australian department of education , science and training ; the national science foundation of china under contract no .
10175071 ; the department of science and technology of india ; the bk21 program of the ministry of education of korea and the chep src program of the korea science and engineering foundation ; the polish state committee for scientific research under contract no .
2p03b 01324 ; the ministry of science and technology of the russian federation ; the ministry of education , science and sport of the republic of slovenia ; the swiss national science foundation ; the national science council and the ministry of education of taiwan ; and the u.s.department of energy .
9 s. j. brodsky and g. p. lepage , phys .
d * 24 * ( 1981 ) 1808 . c. vogt , in proceedings of photon 2000 : ambleside , england , august 26th-31st , american institute of physics ( 2001 ) , 345 .
j. dominick _
et al . _ , cleo collaboration , phys .
d * 50 * ( 1994 ) 3027 .
m. diehl , p. kroll and c. vogt , phys .
b * 532 * ( 2002 ) 99 .
a. heister _
et al . _ ,
aleph collaboration , phys .
b * 569 * ( 2003 ) 140 .
a. abashian _
et al . _ ,
belle collaboration , nucl .
instr . meth .
a * 479 * ( 2002 ) 117 .
s. kurokawa and e. kikutani , nucl .
instr . meth .
a * 499 * ( 2003 ) 1 .
s. jadach and z. was , comp .
. commun . * 85 * ( 1995 ) 453 .
f. a. berends , p. h. daverveldt and r. kleiss , nucl .
b * 253 * ( 1985 ) 441 ; + f. a. berends , p. h. daverveldt and r. kleiss , comp .
* 40 * ( 1986 ) 285 .
et al . _ , phys .
b * 540 * ( 2002 ) 33 .
the detector response is simulated with geant , r. brun _ et al .
_ , geant 3.21 , cern report dd / ee/84 - 1 , 1984 . s. uehara , kek report 96 - 11 ( 1996 ) .
s. jadach , b. f. l. ward and z. was , comp .
* 130 * ( 2000 ) 260 .
s. j. brodsky and g. r. farrar , phys .
d * 11 * ( 1975 ) 1309
_ , belle collaboration , eur . phys .
j. c * 32 * ( 2003 ) 323 .
m. benayoun and v. l. chernyak , nucl .
b * 329 * ( 1990 ) 285 .
s. eidelman _
et al . _ ,
b * 592 * ( 2004 ) 1 . | we have measured @xmath0 and @xmath1 production in two - photon collisions using @xmath2 of data collected with the belle detector at the asymmetric energy @xmath3 collider kekb .
the cross sections are measured to high precision in the two - photon center - of - mass energy ( @xmath4 ) range between @xmath5 and angular region @xmath6 .
the cross section ratio @xmath7 is measured to be @xmath8 in the range of @xmath9 , where the ratio is energy independent .
we observe a @xmath10 behavior of the cross section in the same @xmath4 range .
production of @xmath11 and @xmath12 mesons is observed in both @xmath13 and @xmath14 modes .
belle preprint 2004 - 39 + kek preprint 2004 - 80 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , and two - photon collisions , mesons , qcd , charmonium 12.38qk , 13.25.gv , 13.66.bc , 13.85.lg |
radio continuum emission traces star formation on timescales of up to 100 myr @xcite . two physical processes resulting from massive star formation
produce most of the radio continuum emission between 1 and 100 ghz in star - forming galaxies : ( 1 ) nonthermal synchrotron emission from relativistic electrons accelerated by magnetic fields as a result of recent supernovae and ( 2 ) thermal free - free emission from gas ionized by young massive stars @xcite .
the nonthermal emission is closely tied to the number of supernova - generating massive stars produced in recent episodes of star formation , while the thermal emission gives a nearly direct measure of the current equivalent number of o stars via the ionizing flux in the sampled area . since each component traces a physical process with a well - known timescale , we can use measurements of the radio continuum to determine star formation rates and constrain the ages of recent episodes of star formation .
recent studies of nearby star - forming galaxies with interferometers have emphasized resolving individual star - forming regions ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
since radio continuum emission is not affected by extinction , it can be used to observe deeply embedded regions of current star formation that have not yet shed their surrounding material and are thus invisible at shorter wavelengths .
these studies have taken advantage of interferometers exceptional spatial resolution to probe very young starbursts whose optical emission is obscured by dust . while these studies have been invaluable for determining star formation properties in galaxies outside of our own , the high angular resolution and missing short - spacing data of interferometers , especially at higher frequencies , `` resolves out '' the diffuse radio continuum emission outside of compact star - forming regions .
this effect disproportionally impacts synchrotron emission , which tends to be much more diffuse than the primarily thermal emission surrounding areas of ongoing massive star formation @xcite . unlike interferometers ,
single dish telescopes are not plagued by missing short spacings .
therefore , these telescopes provide a way to simultaneously measure the compact thermal and diffuse non - thermal components of a galaxy s radio continuum emission in order to characterize its _ global _ star formation properties . determining the relative contributions of the thermal and nonthermal components of the measured flux of entire galaxies
can be challenging .
fortunately , each component has a distinct behavior with respect to frequency , and therefore we can model radio continuum emission with a simple two - component fit .
radio continuum flux follows a power law relation such that @xmath2 , where @xmath3 is the spectral index that is characteristic of a source s emission .
optically thin thermal emission exhibits @xmath4 -0.1 , and nonthermal emission exhibits @xmath5 -0.8 @xcite . to determine a reliable fit to these parameters , observations sampling the same physical area at multiple ,
widely - spaced frequencies are required .
if only one frequency is observed , it is impossible to determine the relative contributions of each emission process without previous knowledge of the source .
since single dish telescopes are sensitive to both compact thermal and diffuse nonthermal emission over large spatial extents , they are useful for constraining the large - scale properties of multiple components of a galaxy s radio continuum emission , and are thus powerful probes of star formation .
the goal of this paper is to characterize the _ global _ star formation properties of local galaxies .
our observations were taken in four independent channels continuous in frequency across the full 26 - 40 ghz span of ka band .
this range in frequencies is where a typical star - forming galaxy s global radio continuum emission would be expected to contain relatively equal amounts of flux from steep - spectrum synchrotron and flat - spectrum thermal sources @xcite .
our observations are thus ideal for approximating the relative contributions of each type of emission at this `` lever arm '' frequency range .
using new radio continuum observations centered at 27.75 ghz , 31.25 ghz , 34.75 ghz , and 38.25 ghz , as well as archival nvss 1.4 ghz and iras 60 @xmath6 and 100 @xmath6 data , we have determined these galaxies thermal fractions and star formation rates .
we have also explored the radio - far - infrared correlation in these galaxies and its implications for their star formation timescales .
we will describe the galaxy sample and our observations and data reduction in section 2 , present our results and address the process of fitting spectral energy distributions to our data in section 3 , and finally conclude in section 4 .
we selected a heterogeneous group of 27 local ( d @xmath0 70 mpc ) , well - studied star - forming galaxies with known thermal radio continuum emission .
our sample contains galaxies spanning a variety of shapes , sizes , and environments , from blue compact dwarfs to grand - design spirals , including major and minor mergers , with members of compact groups as well as more isolated galaxies ( see table [ fig : obssummary ] for galaxy types ) .
the intention was to observe as many types of star - forming galaxies as possible to probe star formation in a diverse range of environments .
see table [ fig : obssummary ] and figure [ fig : sampleproperties ] for sample properties . for more information on each galaxy s previous radio continuum observations and discussions of their properties ,
see the papers named in table [ fig : obssummary ] .
the galaxies in our sample span a range of distances ( 1 - 70 mpc ) and properties .
they all have previously detected radio emission and ongoing star formation that covers three orders of magnitude in star formation rate .
thus , they are strong targets for a study of global radio continuum properties at a frequency range that probes both thermal free - free and nonthermal synchrotron star formation indicators .
as seen in figure [ fig : sampleproperties ] , these galaxies are largely less massive and have higher star formation rates than the milky way , and have subsolar metallicities
. however , their properties are not so similar that they can be considered as a single class
. it would not be surprising if their radio continuum properties also encompassed a range of values .
our analysis is best understood as reflecting properties of nearby star - forming galaxies , though it is beyond the scope of this paper to perform detailed analysis on each galaxy individually .
we observed the galaxies in our sample with the caltech continuum backend ( ccb ) on the robert c. byrd green bank telescope ( gbt ) using single pointings .
the ccb is designed for the gbt s dual - beam ka band receiver spanning the entire range of frequencies from 26 - 40 ghz .
the primary observing mode of the ccb is a 70 second `` nod '' , where each beam takes a turn as the on - source beam while the other beam is off - source .
we observed 24 of the galaxies using a single nod each , while we observed eight galaxies using multiple nods , which we then averaged .
five galaxies in our sample were observed on two different nights with both of these methods ; we treated these on a case - by - case basis and chose the observation(s ) with the best weather and elevation conditions .
we used the standard nrao primary flux calibrators 3c 147 and 3c 48 for flux calibration , as well as nearby pointing calibrators to ensure accurate pointing . see table [ fig : obssummary ] for a summary of the observations .
we reduced our data using idl reduction routines developed by b. mason ( for details on the data reduction process , see * ? ? ? * ) .
data with wind speeds over 5 @xmath7 were excluded due to the possibility of large pointing errors .
we detected 22 galaxies in all four channels .
when a galaxy s flux was lower than the 2@xmath8 level in one or more of the four channels , we combined the channels fluxes to produce one average flux across the entire band , centered at 33 ghz .
one galaxy , pox 4 , was detected at the @xmath9 level after averaging the four channels .
three additional galaxies were marginally detected ( between 2@xmath8 and 3@xmath8 ) using this method .
we report an upper limit 33 ghz flux for one galaxy , tol 35 .
the galaxies observed fluxes are reported in table [ fig : uncorrectedflux ] .
since the angular size of a telescope s beam is inversely proportional to the frequency observed , the beam size of the gbt varies appreciably across the 26 - 40 ghz range of our observations ( @xmath10 in the lowest - frequency sub - band versus @xmath11 in the highest - frequency sub - band , see figure [ fig : beamsizes ] for an illustration ) .
we followed the procedure of @xcite to correct for differing beam sizes in each of the four sub - bands .
first , we imaged an archival vla radio continuum map of each galaxy ( typically at frequencies of 4 - 10 ghz ) using the aips task imagr . for these maps , we selected the archival uv data from the nrao science data archive with the closest beamsize to our ka - band data .
we explicitly imposed each of the four ccb beam sizes on these images using bmin and bmaj ( assuming a circular beam ) .
we determined correction factors for each beam by normalizing the flux contained in each beam area in the archival map to the flux in the 34.75 ghz port s @xmath12 beam .
this procedure partially adjusts for more flux being observed at lower frequencies due to these frequencies intrinsically larger beam sizes .
we then could approximate `` beam - matched '' flux measurements to determine spectral indices between 26 - 40 ghz ( see figure [ fig : histogram ] for an illustration of the galaxies spectral indices before and after applying the corrections ) .
ngc 1222 did not have available archival data , so we applied to it the average correction factors of all of the other galaxies .
we emphasize that these correction factors are only approximate . in many cases ,
they are based on resolved archival images that may not contain all of the galaxies radio flux . in particular
, these resolved images may contain most of the thermal emission , which tends to be compact , but underestimate the galaxies nonthermal emission , which tends to be diffuse .
this could bias the correction factors to be closer to 1.0 than should be the case , especially in the highest - frequency ( and thus highest - resolution ) channel .
see table [ fig : beamcorrections ] for the beam correction factors for each galaxy .
the dominant sources of uncertainty in our beam corrections are systematic errors due to the geometries of our sources .
the smallest corrections possible are for a galaxy whose most diffuse , extended flux is still contained within the smallest beam and is unresolved by the lower - frequency interferometric observations .
this type of source would look identical to all four of the gbt beam sizes . in this case
, the correction factors would be 1.0 for each sub - band . for a source much more extended than the beam sizes , the maximum deviations from no beam corrections in each sub - band are -36@xmath13 , -19@xmath13 , 0@xmath13 , and + 21@xmath13 . pointing offsets from the peak of radio emission
can also be sources of systematic error , though the errors depend on whether the source is compact or extended , and the magnitude of the pointing offset from the central peak of radio emission .
these errors are typically smaller than the maximum deviations discussed above .
since we do not know how much diffuse emission is missing in the archival data , we do not have enough information to quantify uncertainties in the beam correction factors for each galaxy .
many of the galaxies that we observed were more extended than the @xmath1 beam size of the gbt at 33 ghz . in these cases , radio continuum fluxes and star formation rates
should only be interpreted as covering the inner @xmath1 of the galaxies .
the galaxies with resolved lower - frequency archival data that was more extended than the beam size are flagged with a `` 1 '' in table [ fig : correctedflux ] .
the ccb has a beam separation of @xmath14 between the `` on '' and `` off '' beams .
m 51 and m 101 are more extended than that separation in both optical images ( see figure [ fig : beamseparation ] ) and in maps of their lower - frequency radio continuum emission @xcite . in these cases
, our flux measurements may be lower than the true amount of flux contained within the beam .
there is likely to be radio continuum emission at the `` off '' positions , which would cause an oversubtraction of flux in the reduction process .
for 22 of the 27 galaxies that we observed , the fluxes we report are the first detections ( either in all four sub - bands or averaged ) at @xmath1533 ghz .
four of the galaxies in our sample were previously observed with the ccb by @xcite , three of which were detected ( m 101 was reported as an upper limit by @xcite and is a @xmath16 marginal detection when the four sub - bands were averaged in our observations ) .
only one galaxy in our sample , tol 35 , was not detected when averaging four sub - bands fluxes .
its @xmath17 upper - limit flux is 0.87 mjy .
this galaxy was observed at a very low elevation ( @xmath18 ) , so it was observed with large atmospheric extinction .
see table [ fig : uncorrectedflux ] for the uncorrected fluxes , and table [ fig : correctedflux ] for the fluxes corrected for the differing beam sizes of each frequency .
we fit a spectral energy distribution ( sed ) for each galaxy that was detected in all four sub - bands using the four ccb fluxes and archival nrao vla sky survey ( nvss ) 1.4 ghz fluxes ( measured with a 45@xmath19 aperture ) .
we assumed a two - component fit of nonthermal emission with a spectral index @xmath20 = -0.8 and thermal emission with a spectral index @xmath21 = -0.1 .
these fits are plotted in figure [ fig : sedpanel ] .
though the spectral index of nonthermal emission can vary ( this phenomenon is described further in section 3.2.1 ) , we used this simple model because we only fit to five data points for each galaxy ; our model did not include enough data to justify additional free parameters .
we do not see evidence of anomalous dust emission in the observed regions of these galaxies ( for explanations of anomalous dust , see * ? ? ?
* ; * ? ? ?
our observations are also at frequencies low enough to have negligible contributions from the low - frequency tail of the dust blackbody .
therefore , we did not include any thermal dust emission in our fits .
our spectra also do not show the inverted structure characteristic of self - absorption or optically thick thermal emission , so we did not include either of these components . from these fits , we determined each galaxy s thermal and nonthermal fluxes at 33 ghz .
none of the galaxies have globally flat spectra indicative of purely thermal emission , nor the inverted spectra seen in some resolved observations of very young , obscured thermal sources .
thermal emission was the primary component at 33 ghz in some galaxies , while others had less prominent or even negligible thermal components in the observed regions .
in contrast to radio continuum studies done at high spatial resolution , our single dish observations detect the diffuse synchrotron emission produced by past supernovae in addition to the strong compact thermal emission from h ii regions , so the spectral indices that we derive are typically much steeper than those derived only from detections of compact radio sources .
since our observations do not spatially separate regions of thermal and nonthermal emission , we can not further distinguish the two components in that way .
the fitted spectra for eight of the 27 galaxies ( arp 217 , ngc 4449 , ngc 2903 , maffei ii , ngc 4038 , m 51 , ngc 4490 , and ngc 1741 ) are significantly steeper than can be fit by a combination of thermal ( @xmath21 = -0.1 ) and nonthermal ( @xmath20 = -0.8 ) components ( see figure [ fig : sedpanel ] ) . when we could not fit a galaxy s sed with both the thermal and nonthermal components at the @xmath22 level , we used only a single - component fit that assumed no thermal flux and a fixed nonthermal spectral index of @xmath20 = -0.8 for consistency .
the thermal fluxes and associated properties of this group of galaxies are reported as upper limits .
we used the total flux in the 34.75 ghz channel plus 3@xmath8 as a conservative upper limit to the thermal flux in these cases .
there are two possible explanations for the steep spectra that we see in some galaxies
. there could be technical considerations due to imperfect beam - matching in our data , or there could be physical processes taking place within these galaxies causing their spectra to steepen at high frequencies . in order to have more accurate sed fits and more precise star formation rates we would need to have beam - matched observations of the same regions at many different frequencies .
the correction factors for differing beam sizes that are given in table [ fig : beamcorrections ] are limited by being calculated using higher - resolution data that could be missing extended emission .
if extended emission is missing in the archival data , the correction factors in table [ fig : beamcorrections ] could be closer to 1.0 than is actually the case . while all of the correction factors calculated act to flatten the sed between 26 ghz and 40 ghz with respect to the uncorrected data , it is possible that they do not flatten the sed enough if they do not reflect contributions from extended emission ( as discussed in section 2.2 ) .
in addition , we did not correct for mismatched beams between the @xmath23 nvss data and the @xmath24 ccb data .
this beam difference only affects resolved galaxies ( those marked with a `` 1 '' in table [ fig : correctedflux ] ) , which comprise @xmath25 of our sample .
it is possible that synchrotron emission is more adversely affected by the differences in beam sizes than thermal emission .
more diffuse synchrotron emission could be undetected at higher frequencies ( and thus smaller beam sizes ) than would be expected for a smooth flux distribution observed with two apertures of different sizes .
if this is the case in our observed regions , it could explain why some of our galaxies spectra steepen at the frequencies we observed .
it is also possible that the choice of where the gbt beams were pointed within a galaxy could affect its fluxes in different beam sizes .
if the beam is not centered on the galaxy ( in the case of unresolved galaxies ) or is not centered on a bright knot of emission ( in the case of resolved galaxies ) , the smaller beams could contain even less flux than would be expected after corretions for the beams areas .
ngc 1741 and ngc 4490 are likely affected by pointing offsets , as seen by comparing the gbt pointing in table [ fig : obssummary ] to previous radio continuum maps in figure 2 of @xcite and figure 4 of @xcite . as described in section 2.2 , pointing offsets from the peak of radio continuum emission result in the need for larger beam correction factors than derived from the archival radio continuum data , the lack of which result in steep spectra at the observed frequencies .
in addition to the technical issue of mismatched beam sizes , there are possible physical explanations for steep spectra in star - forming galaxies .
it is difficult to distinguish between a spectrum with a nonthermal component having @xmath26 coupled with a low thermal fraction from a spectrum with a steeper nonthermal component coupled with a relatively high thermal fraction @xcite .
though the spectral indices that we used are typical values @xcite , they can vary depending on the physical parameters of the observed regions .
thermal emission can have a positive spectral index if the emission regions are optically thick , though we do not see any evidence that this is occuring on the angular scale of our observations .
nonthermal spectral indices can be positive at low frequencies due to synchrotron self - absorption ( which we do not observe ) , or become more negative with increasing frequency and increasing scale height from the disk due to aging cosmic ray electrons losing energy as they propagate outward from their parent supernovae @xcite .
@xcite calculated the timescale for synchrotron losses for cosmic ray electrons in ngc 4214 to be 44 myr at 1.4 ghz and 18 myr at 8.5 ghz .
there is also some evidence of steepening spectra at higher frequencies ( @xmath27 10 ghz ) for luminous and ultra - luminous infrared galaxies , as well as in the post - starburst galaxy ngc 1569 @xcite .
these authors hypothesize that winds or outflows may disperse synchrotron emission from its parent source more quickly than would be expected for simple diffusion .
this rapid dispersal could cause a dearth of synchrotron emission at higher frequencies on shorter timescales than would be predicted from the timescale of energy loss .
@xcite also hypothesize that there could be a modified injection spectrum in galaxies where this is the case .
our sample of galaxies does not contain any lirgs or ulirgs , and we do not see steepening in our measurements of ngc 1569 . we are only observing the inner region of ngc 1569 , while the dispersed synchrotron emission resides in its outer halo , so it is not surprising that we do not observe a steepening spectrum in this galaxy .
we suspect that the steep spectra seen in our sample are primarily a result of imperfect beam matching as discussed above .
this is especially likely to be the case for the galaxies resolved by the gbt at 33 ghz , since these galaxies will have emission that is outside of the view of the gbt beam but is included in the nvss flux . as discussed earlier , the galaxies that appear unresolved in archival maps could still have diffuse synchrotron emission that was not detected in the archival data but that is more extended than the @xmath28 beam at 33 ghz .
five of the eight resolved galaxies that were detected in all sub - bands had steep spectra ( four out of those five are classified as spiral galaxies ) , while only three of the fourteen unresolved galaxies had this feature .
two of these three are classified as sabbc galaxies , while the third is classified as peculiar .
it is possible that these steep - spectrum galaxies contain emission in their spiral arms that is extended with respect to the gbt s smaller beam but is observed in the nvss data . in figure
[ fig : sedpanel ] , most of the galaxies with steep spectra ( and thus single component fits ) also showed the nvss 1.4 ghz data point being located above the best - fit line expected for purely nonthermal emission .
this could be a consequence of the larger beam at 1.4 ghz sampling a larger physical area of emission .
even so , the alternative physical explanations merit consideration , especially in the case of the unresolved galaxies . in figure
[ fig : rad - ir ] , which will be discussed further in section 4.6 , the three unresolved galaxies with steep spectra ( ngc 1741 , ngc 2903 , and arp 217 ) have elevated 1.4 ghz fluxes with respect to what would be expected from the radio - far infrared correlation .
since the 33 ghz fluxes of these galaxies are not similarly elevated with respect to their far - infrared fluxes in figure [ fig : rad - ir ] , their steep radio spectra may indicate an internal physical process that strongly increases the amount of synchrotron emission .
the average thermal fraction fit by two - component models at 33 ghz was 54@xmath13 , with a @xmath29 scatter of 24@xmath13 and a range of 10@xmath13 - 90@xmath13 .
the average is consistent , albeit with large scatter , with the average global thermal fraction at 33 ghz in star - forming galaxies without active galactic nuclei following the relation @xmath30 where @xmath31 is the nonthermal spectral index , @xmath32 is the thermal flux at a given frequency , and s is the total flux at that frequency @xcite .
when a two - component fit was not possible , we report the thermal flux as the corrected flux at 34.75 ghz plus @xmath17 , which gives a very conservative upper limit .
we expect from the galaxies seds that their true thermal fractions are very low at 33 ghz , which we assume in the rest of our analysis .
the large scatter in the thermal fraction is likely a consequence of our heterogeneous galaxy sample ; these galaxies are at different stages of evolution and have different star formation rates , stellar populations , and physical properties @xcite .
some of them may have a very recent ( @xmath0 10 myr ) burst of star formation that produces a large amount of free - free emission that dominates their spectra from 1 - 100 ghz .
others may be in between episodes of very active star formation and instead be experiencing a more quiescent phase , which would result in a relatively low thermal fraction and steepening nonthermal component at 33 ghz due to synchrotron energy losses at high frequencies .
thermal emission traces very recent star formation , since it comes from ionized regions around short - lived , massive stars . for a single starburst , a spectrum showing solely thermal emission requires that too few supernovae have yet occurred to detect their emission .
this would constrain the starburst to be less than @xmath15 6 myr old ( or even younger , depending on the mass and lifetime of the most massive o stars in the starburst ; @xcite find the lifetime of a 120 @xmath33 star to be 3.4 myr ) . a complete absence of thermal flux
implies the absence of enough massive o stars to have detectable free - free emission for a long enough period of time that the emission has dissipated from its parent region .
if this was the case , the starburst is likely at least 30 myr old ( the lifetime of the least massive supernova progenitors ) . on the other hand , nonthermal emission probes star formation on longer timescales ( 30 myr @xmath34 100 myr ) .
it is produced by recent supernovae of stars that can be less massive and have longer lifetimes than the o stars that produce thermal emission ( see figure 9 of * ? ? ?
the presence of nonthermal emission implies that the starburst is at least 6 myr old but younger than the timescale dictated by synchrotron energy loss for the galaxy s magnetic field ( @xmath15 100 myr ) @xcite .
we note that there are limits to the amount of each component that we can detect , so the timescales quoted in the previous paragraph are only approximate . to constrain how much nonthermal emission could be present in a spectrum that appears purely thermal , we generated spectra with varying thermal fractions with fluxes at the same five frequencies as those in our data set ( 1.4 ghz , 27.75 ghz , 31.25 ghz , 34.75 ghz , and 38.25 ghz ) and 10@xmath13 errors on the fluxes .
when these spectra are fit with a two - component model assuming @xmath35 and @xmath31 , nonthermal emission can only be detected in the spectra for thermal fractions less than 97@xmath13 .
this means that the galaxy could have some nonthermal emission ( up to @xmath36 for @xmath37 errors on the fluxes ) , but the emission would be undetectable and thus the starburst would appear younger than it is . similarly , a spectrum could look like it contains no thermal emission while actually containing quite a bit . for the same spectra with @xmath37 errors on the fluxes , thermal fractions of up to @xmath38 resulted in undetectable thermal components .
this means that a galaxy could look like its massive star formation has ceased while still having a small thermal component . for the galaxies in our sample , this picture could be more complicated .
the quoted timescales in this section are for an isolated single starburst .
since our observations measure star formation properties on large angular scales , the galaxies may have multiple overlapping generations of star formation that are not easily separated in time .
we are also sampling different structures and physical scales in each galaxy .
for some galaxies , we are only observing the most central region .
for these galaxies , we may be missing the majority of the ongoing star formation happening in outer regions and spiral arms .
for the more compact galaxies , however , we are likely measuring the entirety of the galaxy s star formation within the gbt beam , so our measurements characterize their global star formation properties . for those galaxies whose seds were fit with thermal components , we used their fluxes at 33 ghz to calculate their thermal luminosities .
we then used those luminosities to calculate the number of ionizing photons responsible for the thermal fluxes seen within the gbt beam following equation 2 in @xcite : @xmath39 where @xmath40 is the number of lyman continuum photons emitted by the region on thermal emission , @xmath41 is the electron temperature , and @xmath42 is the thermal luminosity .
the resulting values are detailed in table [ fig : sfrfractableunres ] ( unresolved galaxies ) and table [ fig : sfrfractableres ] ( resolved galaxies ) .
we used an electron temperature of @xmath43k , as is typical for star - forming regions @xcite , and used @xmath44 as the number of lyman continuum photons emitted by an o7.5v star from table 5 in @xcite .
we report the total number of o7.5v stars in the galaxies that are unresolved by the gbt at 33 ghz , and the number of o7.5v stars per square kiloparsec for the resolved galaxies in tables [ fig : sfrfractableunres ] and [ fig : sfrfractableres ] .
as seen in table [ fig : sfrfractableunres ] , the number of o7.5v stars in each unresolved galaxy varies widely ( log @xmath45 o7.5v stars is between 2.42 and 4.66 ) .
this is likely due to the wide range in the unresolved galaxies overall star formation rates and physical areas observed . since we were able to fit nonthermal components for all of our galaxies , we calculated supernova rates ( @xmath46 ) for each of them following equation 18 in @xcite : @xmath47 where @xmath48 is the nonthermal luminosity .
we report the total supernova rate of the unresolved galaxies in table [ fig : sfrfractableunres ] , while for the resolved galaxies we report the supernova rate per square kiloparsec in table [ fig : sfrfractableres ] .
the supernova rates of the unresolved galaxies vary by three orders of magnitude ( log sne rate between -3.72 and -0.71 ) , which is not surprising given the differences in star formation rates and physical areas sampled .
we calculated massive star formation rates ( sfrs ) from thermal fluxes for each galaxy whose seds have a thermal component and from nonthermal fluxes for all of our galaxies following equations 21 and 23 of @xcite : @xmath49 @xmath50 where @xmath42 and @xmath48 are thermal and nonthermal luminosities , respectively , calculated from each galaxy s thermal and nonthermal fluxes , and @xmath51 ghz .
these equations are derived from equations 2 and 18 of @xcite ( reproduced as equations 2 and 3 in this paper ) .
those equations were derived assuming ( 1 ) an extended miller - scalo imf @xcite with an exponent of @xmath52 , ( 2 ) that all stars with masses greater than @xmath53 become supernovae , and ( 3 ) that dust absorption is negligible @xcite .
we then scaled the massive sfrs generated by each equation by a factor of 5.6 to transform them to total sfrs ( m @xmath54 0.1@xmath33 ) calculated with a kroupa imf @xcite .
the galaxies sfrs calculated from their thermal and nonthermal fluxes are shown in table [ fig : sfrfractableunres ] ( unresolved galaxies ) and table [ fig : sfrfractableres ] ( resolved galaxies ) .
we report the total massive sfrs of the unresolved galaxies , while we report the massive sfr per square kiloparsec of the resolved galaxies .
all of the galaxies for which we calculated both thermal and nonthermal sfrs showed agreement between the two to within an order of magnitude , but not necessarily to within their margins of uncertainty .
the disagreement correlates with the thermal fractions of each galaxy : galaxies with high thermal fractions were likely to have higher thermal sfrs than nonthermal sfrs , while galaxies with low thermal fractions showed the opposite relation . like the differences in thermal fractions between galaxies in our sample
, disagreement could be due to the different star formation timescales traced by the thermal and nonthermal fluxes . since these two emission components are caused by physical processes that operate over differing lengths of time ( as discussed in section 3.3.1 )
, it is possible that the discrepancies between the star formation rates could be used to infer the recent star formation histories of the observed regions .
we compared the radio continuum sfrs to monochromatic sfrs from 24@xmath55 fluxes as described in @xcite .
the galaxies sfrs ( for the unresolved galaxies ) and sfr densities ( for the resolved galaxies ) derived from 24@xmath56 fluxes are listed in table [ fig : sfrfractableunres ] and table [ fig : sfrfractableres ] . in figure
[ fig : comparesfrs ] , we compare the sfrs derived from thermal and nonthermal radio continuum fluxes of the unresolved galaxies for which we fit two - component seds to sfrs derived from 24@xmath56 fluxes .
we find that most of the galaxies in our sample have higher radio continuum sfrs ( both from thermal and nonthermal fluxes ) than sfrs from 24@xmath56 data .
one possible explanation for this is that extinction is lower at radio wavelengths than it is at 24@xmath55 .
another possible explanation is that since radio continuum emission traces very young star formation while 24@xmath56 emission traces less recent star formation , higher sfrs calculated from radio continuum observations than from 24@xmath56 data could be another indication that our sample of galaxies is undergoing recent star formation .
there is a well - established tight correlation between far - infrared ( fir ) and radio flux in star - forming galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
when plotted on a log - log scale , the relationship between radio continuum and fir flux for star - forming galaxies appears linear .
this correlation has been well - studied at low frequencies ( @xmath15 1.4 ghz ) where synchrotron emission is the dominant component of radio emission in a star - forming galaxy .
we investigated whether this correlation could also be found on a global scale at 33 ghz , where synchrotron emission is weaker than it is at 1.4 ghz and the relative contribution from thermal emission is more significant .
we limited our study of the radio - fir correlation to the galaxies in our sample that are unresolved with the gbt beam at 33 ghz ( as discussed in section 2.2 ) .
we chose this limit to ensure that we were observing both the total area of radio emission and total area of far - infrared emission in each galaxy .
this minimizes issues related to the different beam sizes of the gbt and iras ( objects are considered point sources to iras if they are more compact than 1@xmath57 at 60 @xmath6 and 2@xmath57 at 100 @xmath6 ) .
we fit a power law to our 33 ghz flux as a function of total fir flux .
the total fir flux was determined by a combination of archival iras 100 @xmath6 and 60 @xmath6 fluxes as described in @xcite ( @xmath58 ) .
we chose to compute each galaxy s 33 ghz flux by taking the average of its fluxes in the four sub - bands .
we used this measure ( rather than the flux at 33 ghz inferred from the galaxies seds ) in order to eliminate possible uncertainties in the flux due to using assumed spectral indices in our fits .
we found that the fluxes were related by @xmath59 .
this correlation is relatively well - fit ( the fractional errors of both fit parameters are small ) even though our sample contains a wide range of thermal fractions .
@xcite found a similar correlation between 33 ghz and @xmath60 fluxes for resolved nuclei and individual star - forming regions of galaxies .
we find that the radio - fir correlation at 33 ghz can be extended to global measurements of galaxies fluxes . as a control of the tightness of the radio - fir correlation in our sample
, we also fit a relationship between the galaxies nvss 1.4 ghz fluxes and their total fir fluxes .
this relationship for the unresolved galaxies in our sample is @xmath61 .
the fractional uncertainties on the fit parameters are similar to those of the fit at 33 ghz .
we plot both correlations in figure [ fig : rad - ir ] .
as discussed in section 3.2 , we have determined thermal fractions from sed fits assuming fixed thermal and nonthermal spectral indices .
due to the limited number of radio data points we have for each galaxy , we can not more accurately constrain the thermal fractions at 33 ghz of the galaxies in our sample at this time . therefore
, we do not have enough information to definitively isolate thermal and nonthermal components to explore whether the radio - fir correlation is equally tight for each . as an estimate , we have coded approximate thermal fractions in the plot .
even given these limitations , we are confident that a correlation exists between the unresolved galaxies total radio flux at 33 ghz and total fir flux . @xcite found a similar correlation at 33 ghz for resolved nuclei and star - forming regions of galaxies . to further constrain the radio - fir correlation at 33 ghz in our sample , we calculated @xmath62 for each galaxy
. @xmath62 is a logarithmic measure of the ratio of total far - infrared flux ( @xmath63 in janskys ) to radio continuum flux ( @xmath64 ) in units of @xmath65 at a given frequency .
it is defined in @xcite as @xmath66 the average @xmath67 for our sample is @xmath68 , with a 1@xmath8 scatter of 0.3 .
@xcite reported that at 1.4 ghz , the average value of @xmath69 from a large sample of galaxies is @xmath70 .
the average value of @xmath62 at 1.4 ghz for this set of galaxies is @xmath71 , consistent with the @xcite value .
since @xmath62 is a function of the ratio of fir flux to radio flux at a given frequency , it makes sense that @xmath62 is larger using 33 ghz fluxes than it is using 1.4 ghz fluxes ( star - forming galaxies are generally much brighter at 1.4 ghz than at 33 ghz ) .
the scatter on @xmath62 at 33 ghz is larger than that at 1.4 ghz , which indicates that the radio - fir correlation is not as tight at 33 ghz as at 1.4 ghz .
this may be due to contamination from increased thermal flux at 33 ghz .
if the correlation is solely between synchrotron and fir emission , thermal flux at 33 ghz will increase the scatter in the correlation .
however , due to our small sample size , we can not rule out the possibility that the correlation is just as strong at 33 ghz , where thermal fractions are higher , as it is at 1.4 ghz , where nonthermal emission is typically much stronger .
we note that the galaxies with the highest thermal fractions lie above the fitted correlation at 33 ghz , while the same is not true at 1.4 ghz , which supports thermal emission being the cause of increased scatter .
in addition to plotting the radio - fir correlation , we also plot the ratio of 33 ghz flux to fir flux , @xmath72 , against @xmath73 for our unresolved galaxies in figure [ fig : inverseqvsalpha ] , similar to @xcite .
the plot shows an increasing @xmath72 for flatter values of @xmath73 .
flatter @xmath73 values are presumably indicative of a higher proportion of thermal flux to nonthermal flux , which is reflected in the highest thermal fractions in our sample also having the flattest @xmath73 .
a correlation between an elevated @xmath72 and flat values of @xmath73 is not surprising if the radio - fir correlation is solely dependent on synchrotron emission .
if the radio - fir correlation was independent of the type of radio emission , @xmath72 should be relatively constant between galaxies and should not be affected by different spectral indices or thermal fractions .
our data support that the radio - fir correlation is independent of a galaxy s thermal emission since the addition of thermal emission results in elevated ratios of 33 ghz flux to fir flux .
when the timescales of the emission mechanisms for thermal , nonthermal , and fir fluxes are taken into account , the observed relationship between the ratio of 33 ghz and fir fluxes and @xmath73 may be a way to age - date an episode of star formation . since thermal flux is only produced by the shortest - lived ( @xmath74 myr ) massive stars ,
its presence in large quantities relative to synchrotron emission is indicative of very young star formation . since in addition to massive stars
, infrared emission also traces less massive stars ( @xmath75 ) that live longer than the @xmath76 stars that produce thermal and nonthermal radio emission @xcite , fir emission is a tracer of star formation on longer timescales .
stars with these masses can live up to @xmath15100 myr , while nonthermal radio emission traces stars with lifetimes of up to @xmath1530 myr and whose emission is detectable for up to 100 myr ( for an illustrative plot of stellar lifetimes , see figure 3 of * ? ? ?
in addition , infrared emission also contains a component from diffuse dust that is heated by lower - mass stars with lifetimes longer than 100 myr .
these timescales could mean that the galaxies that show both flat spectral indices and enhanced @xmath72 also host the youngest areas of ongoing star formation .
this correlation could then be a method of determining approximate ages for galaxies global star formation . as a simple test
, we used a starburst 99 model of a single instantaneous burst using default inputs ( solar metallicity , a 2-component kroupa imf , and no effects of cosmic ray aging , escape , or absorption taken into account ) run for 100 myr @xcite .
this model , depicted in figure [ fig : sb99 ] , shows the flattest spectral indices and highest thermal fractions at the earliest times of the starburst .
similarly , the steepest spectral indices and lowest thermal fractions were seen as the lowest - mass stars that produce supernovae were dying ( at @xmath1540 myr ) .
the starburst s ratio of 33 ghz luminosity to fir luminosity was also high at early times ( between 3 myr and 40 myr ) while the lowest ratios of 33 ghz luminosity to fir luminosity were seen even later ( after 40 myr ) .
while modeling a more robust quantitative relationship between this observed correlation and the age of each galaxy s star - forming episode is beyond the scope of this work ( the simple model we used does not take into account multiple co - existing generations of star formation ) , the apparent relationship between enhanced 33 ghz flux , flat spectral indices , and high thermal fractions is a promising metric for future global radio and far - infrared photometric studies of star - forming galaxies .
our simple model is not robust enough to constrain the timescales uncertainties , but is only meant to be illustrative of a correlation visible in our data .
we have observed 27 local , well - studied , star - forming galaxies between 26 - 40 ghz with the gbt and obtained the first detections at this frequency range for 22 of the galaxies .
we determined the contributions of thermal free - free and nonthermal synchrotron emission to the galaxies total radio emission .
we have used these measures to derive the number of massive , short - lived o stars and the number of recent supernovae in the observed regions of each galaxy .
in addition , we have calculated sfrs for each galaxy using thermal and nonthermal fluxes and explored the radio - fir correlation for the unresolved galaxies .
we found that * none of the galaxies have spectral incides indicative of purely thermal emission ; eight galaxies show spectra that are too steep to fit thermal components , * thermal fractions range from 10@xmath13 to 90@xmath13 , with a median of @xmath77 , * the radio - far infrared correlation holds for the unresolved galaxies at 1.4 ghz and 33 ghz , though the scatter at 33 ghz is larger due to the increased influence of thermal emission at higher frequencies , and * galaxies with flat @xmath73 and high thermal fractions have enhanced radio flux at 33 ghz with respect to far - infrared flux , which identifies them as galaxies with recent star formation .
this is consistent with a simple model of a single starburst .
we found that the observed regions of our galaxies had a diverse mix of radio continuum characteristics , with some galaxies seds being dominated at 33 ghz by the thermal emission indicative of ongoing massive star formation , while others have little or no detectable thermal emission .
even with this spread in the relative contributions of thermal and nonthermal emission , we saw that there is still a correlation between the global 33 ghz and far - infrared flux in the unresolved galaxies .
the scatter in the correlation is larger than that at 1.4 ghz , likely due to the increased influence of thermal emission at 33 ghz .
we can not , however , rule out that the radio - fir correlation is not solely dependent on synchrotron emission .
we also found that higher ratios of 33 ghz emission to fir emission correlated with flatter spectral indices ( and higher thermal fractions ) for unresolved galaxies , which is consistent with younger ages in simple starburst models .
this correlation may be useful as a rough indicator of the age of the most recent episode of star formation .
future global studies of more homogeneous galaxy populations or resolved studies of individual star - forming regions will enable better modeling of star formation timescales using this metric .
in giving a broad measure of nearby galaxies radio continuum emission , our observations complement previous studies done with interferometers in which individual star - forming regions in local galaxies were highly resolved . with the gbt , we can simultaneously observe compact and diffuse thermal and nonthermal emission and determine their relative intensities , and in doing so estimate the timescale for the current episode of star formation .
unfortunately , we can not make stricter timescale estimates than those discussed in section 3.3 at this time , as we do not have enough radio data points to robustly fit thermal and nonthermal flux components with varying spectral indices .
obtaining more unresolved radio fluxes at lower and higher frequencies would help this effort .
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 use of nasa s skyview facility ( http://skyview.gsfc.nasa.gov ) located at nasa goddard space flight center .
we thank the telescope operators and support staff at the gbt for assistance with this project . k.r .
acknowledges support from an nrao student observing support award ( gssp10 - 0002 ) .
also thanks brian mason for his help with understanding the ccb observation and data reduction process .
aversa , a. g. , johnson , k. e. , brogan , c. l. , goss , w. m. , & pisano , d. j. 2011 , , 141 , 125 beck , s. c. , turner , j. l. , & kovo , o. 2000 , , 120 , 244 beck , s. c. , turner , j. l. , langland - shula , l. e. , et al .
2002 , , 124 , 2516 beck , s. c. , turner , j. l. , & kloosterman , j. 2007 , , 134 , 1237 bell , e. f. , & de jong , r. s. 2001 , , 550 , 212 beswick , r. j. , pedlar , a. , clemens , m. s. , & alexander , p. 2003
, , 346 , 424 calzetti , d. , wu , s .- y . , hong , s. , et al .
2010 , , 714 , 1256 carlstrom , j. e. , & kronberg , p. p. 1991
, , 366 , 422 chyy , k. t. , & beck , r. 2004 , , 417 , 541 clemens , m. s. , vega , o. , bressan , a. , et al .
2008 , , 477 , 95 clemens , m. s. , scaife , a. , vega , o. , & bressan , a. 2010 , , 405 , 887 condon , j. j. 1992 , , 30 , 575 condon , j. j. , & yin , q. f. 1990 , , 357 , 97 devereux , n. a. , & young , j. s. 1990 , , 359 , 42 draine , b. t. , & lazarian , a. 1998 , , 508 , 157 dumas , g. , schinnerer , e. , tabatabaei , f. s. , et al .
2011 , , 141 , 41 graeve , r. , klein , u. , & wielebinski , r. 1990 , , 238 , 39 heesen , v. , beck , r. , krause , m. , & dettmar , r .- j .
2009 , , 494 , 563 helou , g. , soifer , b. t. , & rowan - robinson , m. 1985 , , 298 , l7 helou , g. , khan , i. r. , malek , l. , & boehmer , l. 1988 , , 68 , 151 israel , f. p. , & de bruyn , a. g. 1988 , , 198 , 109 johnson , k. e. , indebetouw , r. , & pisano , d. j. 2003 , , 126 , 101 johnson , k. e. , & kobulnicky , h. a. 2003 , , 597 , 923 johnson , k. e. , indebetouw , r. , watson , c. , & kobulnicky , h. a. 2004 , , 128 , 610 johnson , k. e. , hunt , l. k. , & reines , a. e. 2009 , , 137 , 3788 kepley , a. a. , zweibel , e. g. , wilcots , e. m. , johnson , k. e. , & robishaw , t. 2011b , , 736 , 139 klein , u. , wielebinski , r. , & beck , r. 1984 , , 135 , 213 kodilkar , j. , kantharia , n. g. , & ananthakrishnan , s. 2011 , , 416 , 522 kroupa , p. 2001
, , 322 , 231 leitherer , c. , schaerer , d. , goldader , j. d. , et al . 1999 , , 123 , 3 leitherer , c. , ortiz otlvaro , p. a. , bresolin , f. , et al .
2010 , , 189 , 309 leroy , a. k. , evans , a. s. , momjian , e. , et al .
2011 , , 739 , l25 lisenfeld , u. , wilding , t. w. , pooley , g. g. , & alexander , p. 2004
, , 349 , 1335 maeder , a. , & meynet , g. 1989 , , 210 , 155 mason , b. s. , weintraub , l. , sievers , j. , et al .
2009 , , 704 , 1433 miller , g. e. , & scalo , j. m. 1979 , , 41 , 513 murphy , e. j. , helou , g. , braun , r. , et al .
2006 , , 651 , l111 murphy , e. j. , helou , g. , condon , j. j. , et al .
2010 , , 709 , l108 murphy , e. j. , bremseth , j. , mason , b. s. , et al .
2012 , arxiv:1210.3360 reines , a. e. , johnson , k. e. , & goss , w. m. 2008 , , 135 , 2222 romano , d. , chiappini , c. , matteucci , f. , & tosi , m. 2005 , , 430 , 491 rosa - gonzlez , d. 2005 , , 364 , 1304 seaquist , e. r. , bell , m. b. , & bignell , r. c. 1985 , , 294 , 546 tovmassian , h. m. 1972 , , 77 , 705 tremonti , c. a. , heckman , t. m. , kauffmann , g. , et al .
2004 , , 613 , 898 tsai , c .- w . ,
turner , j. l. , beck , s. c. , et al .
2006 , , 132 , 2383 turner , j. l. , & ho , p. t. p. 1994
, , 421 , 122 vacca , w. d. , garmany , c. d. , & shull , j. m. 1996 , , 460 , 914 vzquez , g. a. , & leitherer , c. 2005 , , 621 , 695 l l l l l l l c l ngc 520 & 01:24:34.8 & + 03:47:29 & 30.5 @xmath78 2.1 & 122.2 @xmath78 3.2 & 1.7 & sa & 2 & b03 + maffei ii & 02:41:55.9 & + 59:36:14 & 3.11 @xmath78 0.23 & 20.8 @xmath78 4.3 & 0.17 & sbbc & 2 & t94 , t06 + ngc 1222 & 03:08:56.8 & -02:57:18 & 32.4 @xmath78 2.3 & 47.5 @xmath78 3.6 & 1.8 & e / s0 & 1 & b07 + sbs 0335 - 052 & 03:37:44.0 & -05:02:40 & 53.7 @xmath78 3.8 & 13.8 & 3.0 & bcg / starburst & 2 & j09 + ic 342 & 03:46:48.5 & + 68:05:46 & 3.93 @xmath78 0.27 & 599 @xmath78 3.1 & 0.22 & sabc & 1 & m12 , t06 + ngc 1569 & 04:30:49.3 & + 64:50:52 & 1.39 @xmath78 0.11 & 116.7 @xmath78 3.2 & 0.08 & ib & 3 & l04 + vii zw 19 & 04:40:47.3 & + 67:44:09 & 67.4 @xmath78 5.5 & 12.8 @xmath78 3.6 & 3.8 & e & 3 & b00 + ngc 1741 & 05:01:38.3 & -04:15:24 & 54.6 @xmath78 3.8 & 36.1 @xmath78 3.6 & 3.0 & sm & 1 & b00 + ii zw 40 & 05:55:42.6 & + 03:23:32 & 11.1 @xmath78 0.80 & 16.8 & 0.62 & sbc & 3 & b02 + mrk 8 & 07:29:26.3 & + 72:07:44 & 52.5 @xmath78 3.7 & 26.1 @xmath78 4.1 & 2.9 & sbc & 1 & b00 + mrk 86 & 08:13:14.6 & + 45:59:29 & 7.94 @xmath78 1.5 & 61.3 @xmath78 3.4 & 0.44 & sbm & 1 & a11 + ngc 2903 & 09:32:09.7 & + 21:30:02 & 7.39 @xmath78 0.52 & 360.8 @xmath78 3.1 & 0.41 & sabb & 1 & t06 + ngc 2997 & 09:45:38.7 & -31:11:25 & 13.8 @xmath78 0.90 & 307.0 @xmath78 3.1 & 0.77 & sabc & 1 & k11 + mrk 1236 & 09:49:54.1 & + 00:36:58 & 27.6 @xmath78 5.1 & 36.1 @xmath78 3.3 & 1.5 & wr / hii & 1 & b00 + ngc 3077 & 10:03:20.2 & + 68:44:01 & 2.55 @xmath78 0.19 & 157.4 @xmath78 3.2 & 0.14 & s ? & 1 & m12 , r05 + ngc 3125 & 10:06:33.6 & -29:56:08 & 13.8 @xmath78 1.0 & 36.1 @xmath78 3.3 & 0.77 & e & 1 & a11 + arp 233 & 10:32:31.1 & + 54:24:04 & 25.5
@xmath78 1.8 & 34.4 @xmath78 3.4 & 1.4 & i & 1 & t72 , b00 , a11 + arp 217 & 10:38:45.9 & + 53:30:11 & 19.2 @xmath78 1.3 & 57.2 @xmath78 3.3 & 1.1 & sabb & 1 & a11 + haro 3 & 10:45:22.4 & + 55:57:37 & 18.5 @xmath78 1.3 & 40.5 @xmath78 3.4 & 1.0 & sb & 1 & t72 , j04 , a11 + pox 4 & 11:51:11.7 & -20:35:57 & 52.5 @xmath78 3.9 & 19.4 @xmath78 3.4 & 2.9 & sm & 1 & b00 + ngc 4038 & 12:01:52.5 & -18:52:02 & 21.5 @xmath78 1.6 & 161.1 @xmath78 3.2 & 1.2 & sbm & 1 & c04 + ngc 4214 & 12:15:39.2 & + 36:19:41 & 2.94 @xmath78 0.27 & 202.8 @xmath78 3.2 & 0.16 &
ib & 1 & b00 + ngc 4449 & 12:28:10.1 & + 44:05:33 & 3.54 @xmath78 0.25 & 140.3 @xmath78 3.2 & 0.20 & ib & 1 & r08 + ngc 4490 & 12:30:36.7 & + 41:38:26 & 9.22 @xmath78 0.65 & 202.8 @xmath78 3.3 & 0.51 & sbcd & 1 & a11 + tol 35 & 13:27:07.2 & -27:57:26 & 25.2 @xmath78 1.8 & 43.4 @xmath78 3.4 & 1.4 & sab & 1 & b00 + m 51 & 13:29:52.4 & + 47:11:40 & 7.90 @xmath78 0.85 & 414.1 @xmath78 3.1 & 0.44 & sbc & 1 & m12 , t94 , d11 + m 101 & 14:03:12.5 & + 54:20:53 & 6.46 @xmath78 0.18 & 719.7 @xmath78 3.1 & 0.36 & sabc & 1 & m12 , t94 + [ fig : obssummary ]
l l l l l ngc 520 & @xmath79 & @xmath80 & @xmath81 & @xmath82 + maffei ii & @xmath83 & @xmath84 & @xmath85 & @xmath86 + ngc 1222 & @xmath87 & @xmath88 & @xmath89 & @xmath90 + sbs 0335 - 052 & @xmath91 & @xmath92 & @xmath93 & @xmath94 + ic 342 & @xmath95 & @xmath96 & @xmath97 & @xmath98 + ngc 1569 & @xmath99 & @xmath100 & @xmath101 & @xmath102 + vii zw 19 & @xmath103 & @xmath104 & @xmath105 & @xmath106 + ngc 1741 & @xmath107 & @xmath108 & @xmath109 & @xmath110 + ii zw 40 & @xmath111 & @xmath112 & @xmath113 & @xmath114 + mrk 8 & @xmath115 & @xmath116 & @xmath117 & @xmath118 + ngc 2903 & @xmath119 & @xmath120 & @xmath121 & @xmath122 + ngc 2997 & @xmath123 & @xmath124 & @xmath125 & @xmath126 + ngc 3077 & @xmath127 & @xmath128 & @xmath129 & @xmath130 + ngc 3125 & @xmath131 & @xmath132 & @xmath133 & @xmath134 + arp 233 & @xmath135 & @xmath136 & @xmath137 & @xmath138 + arp 217 & @xmath139 & @xmath140 & @xmath141 & @xmath142 + haro 3 & @xmath143 & @xmath144 & @xmath145 & @xmath146 + ngc 4038 & @xmath147 & @xmath148 & @xmath149 & @xmath150 + ngc 4214 & @xmath151 & @xmath152 & @xmath153 & @xmath154 + ngc 4449 & @xmath155 & @xmath156 & @xmath157 & @xmath158 + ngc 4490 & @xmath159 & @xmath160 & @xmath161 & @xmath162 + m 51 & @xmath163 & @xmath164 & @xmath165 & @xmath166 + mrk 86 & 0.42 @xmath78 0.19 & & & + mrk 1236 & 0.99 @xmath78 0.40 & & & + pox 4 & 1.62 @xmath78 0.29 & & & + tol 35 & @xmath167 & & & + m 101 & 0.69 @xmath78 0.27 & & & + ngc 520 & 0.99 & 0.99 & 1 & 1.01 + maffei ii & 0.89 & 0.95 & 1 & 1.05 + ngc 1222 & & & & + sbs 0335 - 052 & 0.98 & 0.99 & 1 & 1.01 + ic 342 & 0.93 & 0.97 & 1 & 1.03 + ngc 1569 & 0.81 & 0.91 & 1 & 1.09 + vii zw 19 & 0.97 & 0.99 & 1 & 1.02 + ngc 1741 & 0.98 & 0.99 & 1 & 1.01 + ii zw 40 & 0.95 & 0.97 & 1 & 1.03 + mrk 8 & 0.92 & 0.96 & 1 & 1.04 + ngc 2903 & 0.96 & 0.98 & 1 & 1.03 + ngc 2997 & 0.89 & 0.94 & 1 & 1.07 + ngc 3077 & 0.97 & 0.98 & 1 & 1.02 + ngc 3125 & 0.92 & 0.96 & 1 & 1.04 + arp 233 & 0.98 & 0.99 & 1 & 1.01 + arp 217 & 0.87 & 0.93 & 1 & 1.07 + haro 3 & 0.95 & 0.98 & 1 & 1.03 + ngc 4038 & 0.83 & 0.94 & 1 & 1.03 + ngc 4214 & 0.87 & 0.94 & 1 & 1.07 + ngc 4449 & 0.78 & 0.89 & 1 & 1.11 + ngc 4490 & 0.75 & 0.87 & 1 & 1.13 + m 51 & 0.83 & 0.91 & 1 & 1.08 + average & 0.91 & 0.95 & 1 & 1.05 + standard dev & 0.07 & 0.04 & 0 & 0.04 + [ fig : beamcorrections ] l l l l l l l ngc 520 & 21.54 @xmath78 0.39 & 19.30 @xmath78 0.19 & 17.44 @xmath78 0.27 & 15.58 @xmath78 0.43 & -1.01 @xmath78 0.10 & + maffei ii & 20.62 @xmath78 0.28 & 18.45 @xmath78 0.16 & 16.36 @xmath78 0.20 & 14.68 @xmath78 0.29 & -1.06 @xmath78 0.07 & 1 + ngc 1222 & 8.94 @xmath78 0.53 & 8.14 @xmath78 0.25 & 7.91 @xmath78 0.36 & 7.62 @xmath78 0.54 & -0.50 @xmath78 0.29 & 2 + sbs 0335 - 052 & 0.65 @xmath78 0.20 & 0.61 @xmath78 0.12 & 0.69 @xmath78 0.18 & 0.53 @xmath78 0.24 & -0.65 @xmath78 1.7 + ic 342 & 33.10 @xmath78 3.5 & 30.10 @xmath78 0.87 & 27.82 @xmath78 1.1 & 26.01 @xmath78 2.2 & -0.75 @xmath78 0.42 & + ngc 1569 & 23.17 @xmath78 0.19 & 22.45 @xmath78 0.12 & 21.31 @xmath78 0.15 & 20.44 @xmath78 0.21 & -0.39 @xmath78 0.04 & 1 + vii zw 19 & 3.07 @xmath78 0.27 & 2.54 @xmath78 0.14 & 2.26 @xmath78 0.19 & 2.15 @xmath78 0.27 & -1.10 @xmath78 0.48 + ngc 1741 & 2.08 @xmath78 0.31 & 2.05 @xmath78 0.18 & 1.67 @xmath78 0.22 & 0.85 @xmath78 0.32 & -2.78 @xmath78 1.3 & + ii zw 40 & 14.33 @xmath78 2.0 & 13.70 @xmath78 0.50 & 12.99 @xmath78 0.62 & 12.59 @xmath78 1.3 & -0.41 @xmath78 0.54 + mrk 8 & 2.97 @xmath78 0.38 & 2.72 @xmath78 0.39 & 2.21@xmath78 0.33 & 2.63@xmath78 0.42 & -0.38 @xmath78 0.64 & + ngc 2903 & 13.91 @xmath78 1.5 & 12.54 @xmath78 0.52 & 11.15 @xmath78 0.82 & 9.89 @xmath78 1.2 & -1.06 @xmath78 0.51 & + ngc 2997 & 4.58 @xmath78 1.1 & 4.48 @xmath78 0.45 & 4.38 @xmath78 0.63 & 4.17 @xmath78 0.85 & -0.29 @xmath78 0.98 & 1 + ngc 3077 & 6.73 @xmath78 0.39 & 5.81 @xmath78 0.38 & 5.45 @xmath78 0.32 & 4.80 @xmath78 0.43 & -1.05 @xmath78 0.33 & + ngc 3125 & 7.49 @xmath78 1.0 & 7.13 @xmath78 0.44 & 6.52 @xmath78 0.58 & 6.15 @xmath78 0.79 & -0.62 @xmath78 0.58 & + arp 233 & 4.14 @xmath78 0.43 & 4.01 @xmath78 0.40 & 3.49 @xmath78 0.31 & 2.96 @xmath78 0.43 & -1.05 @xmath78 0.56 & + arp 217 & 22.60 @xmath78 0.43 & 20.14 @xmath78 0.40 & 19.04 @xmath78 0.31 & 17.40 @xmath78 0.43 & -0.81 @xmath78 0.10 & + haro 3 & 5.79 @xmath78 0.43 & 5.38 @xmath78 0.40 & 4.78 @xmath78 0.31 & 4.48 @xmath78 0.43 & -0.80 @xmath78 0.38 + ngc 4038 & 6.45 @xmath78 0.96 & 5.11 @xmath78 0.42 & 4.64 @xmath78 0.55 & 4.39 @xmath78 0.74 & -1.20 @xmath78 0.70 & 1 + ngc 4214 & 6.25 @xmath78 0.93 & 5.84 @xmath78 0.38 & 5.26 @xmath78 0.48 & 5.01 @xmath78 0.69 & -0.69 @xmath78 0.63 & 1 + ngc 4449 & 3.41 @xmath78 0.90 & 3.07 @xmath78 0.38 & 2.73 @xmath78 0.44 & 2.56 @xmath78 0.65 & -0.90 @xmath78 1.1 & 1 + ngc 4490 & 2.55 @xmath78 0.90 & 2.30 @xmath78 0.38 & 1.64 @xmath78 0.46 & 1.50 @xmath78 0.66 & -1.65 @xmath78 1.8 & 1 + m 51 & 6.36 @xmath78 0.94 & 5.32 @xmath78 0.38 & 4.89 @xmath78 0.47 & 3.94 @xmath78 0.65 & -1.49 @xmath78 0.69 & 1,3 + l l l l l l l ngc 520 & 0.24 @xmath78 0.01 & 4.65 & -0.71 & 7.16 @xmath78 0.43 & 26.7 @xmath78 3.7 & 5.99 @xmath78 0.94 + ngc 1222 & 0.43 @xmath78 0.02 & 4.59 & -1.14 & 6.26 @xmath78 0.48 & 9.8 @xmath78 1.4 & 5.03 @xmath78 0.79 + sbs0335 - 052 & 0.79 @xmath78 0.06 & 4.18 & -2.25 & 2.43 @xmath78 0.50 & 0.77 @xmath78 0.25 & + ic 342 & 0.53 @xmath78 0.02 & 3.41 & -2.50 & 0.41 @xmath78 0.03 & 0.43 @xmath78 0.06 & 0.60 @xmath78 0.09 + vii zw 19 & 0.38 @xmath78 0.003 & 4.66 & -0.97 & 7.28 @xmath78 0.07 & 14.4 @xmath78 2.3 & 16.6 @xmath78 3.0 + ngc 1741 & @xmath0 1.00 & @xmath0 4.87 & -1.03 & @xmath0 12.0 & 12.7 @xmath78 1.8 & 3.80 @xmath78 0.62 + ii zw 40 & 0.91 @xmath78 0.01 & 4.20 & -2.63 & 2.57 @xmath78 0.09 & 0.32 @xmath78 0.05 & 0.46 @xmath78 0.08 + mrk 8 & 0.50 @xmath78 0.05 & 4.58 & -1.26 & 6.1 @xmath78 1.1 & 7.4 @xmath78 1.2 & 2.19 @xmath78 0.38 + ngc 2903 & @xmath0 0.94 & @xmath0 3.90 & -1.92 & @xmath0 1.28 & 1.62 @xmath78 0.23 & 0.31 @xmath78 0.05 + ngc 3077 & 0.66 @xmath78 0.02 & 2.42 & -3.72 & 0.042 @xmath78 0.003 & 0.026 @xmath78 0.004 & 0.0022 @xmath78 0.004 + ngc 3125 & 0.77 @xmath78 0.02 & 4.03 & -2.35 & 1.73 @xmath78 0.11 & 0.61 @xmath78 0.10 & 0.30 @xmath78 0.06 + arp 233 & 0.71 @xmath78 0.02 & 4.25 & -1.99 & 2.89 @xmath78 0.22 & 1.40 @xmath78 0.21 & 1.29 @xmath78 0.21 + arp 217 & @xmath0 0.99 & @xmath0 4.92 & -0.92 & @xmath0 13.4 & 16.2 @xmath78 2.1 & 3.71 @xmath78 0.61 + haro 3 & 0.85 @xmath78 0.02 & 4.19 & -2.40 & 2.51 @xmath78 0.13 & 0.54 @xmath78 0.10 & 0.65 @xmath78 0.11 + mrk 86 & & @xmath0 2.82 & @xmath0 -3.02 & @xmath0 0.11 & @xmath0 0.13 & 0.037 @xmath78 0.015 + mrk 1236 & & @xmath0 4.25 & @xmath0 -1.60 & @xmath0 2.88 & @xmath0 3.43 & 0.36 @xmath78 0.15 + pox 4 & & @xmath0 4.87 & @xmath0 -0.98 & @xmath0 11.8 & @xmath0 14.1 & 0.87 @xmath78 0.27 + tol 35 & & @xmath0 3.99 & @xmath0 -1.86 & @xmath0 1.57 & @xmath0 1.87 & + l l l l l l l maffeiii & @xmath0 0.97 & @xmath0 4.27 & -1.57 & @xmath0 2.99 & 3.69 @xmath78 0.77 & 0.039 @xmath78 0.009 + ngc 1569 & 0.11 @xmath78 0.03 & 3.41 & -1.52 & 0.41 @xmath78 0.15 & 4.06 @xmath78 0.92 & 0.31 @xmath78 0.07 + ngc 2997 & 0.32 @xmath78 0.01 & 3.19 & -2.34 & 0.25 @xmath78 0.04 & 0.63 @xmath78 0.12 & + ngc 4038 & @xmath0 0.87 & @xmath0 3.84 & -1.95 & @xmath0 1.11 & 1.53 @xmath78 0.33 & + ngc 4214 & 0.58 @xmath78 0.03 & 3.55 & -2.44 & 0.56 @xmath78 0.12 & 0.49 @xmath78 0.13 & 0.076 @xmath78 0.022 + ngc 4449 & @xmath0 1.00 & @xmath0 3.65 & -2.25 & @xmath0 0.72 & 0.76 @xmath78 0.15 & + ngc 4490 & @xmath0 0.92 & @xmath0 3.52 & -2.29 & @xmath0 0.53 & 0.69 @xmath78 0.15 & 0.16 @xmath78 0.04 + m 51 & @xmath0 0.94 & @xmath0 3.84 & -1.98 & @xmath0 1.11 & 1.41 @xmath78 0.43 & 0.13 @xmath78 0.04 + m 101 & & @xmath0 2.83 & @xmath0 -3.02 & @xmath0 0.26 & @xmath0 0.32 & 0.015 @xmath78 0.002 + l
l l l l l ngc 520 & @xmath168 & @xmath169 & @xmath170 & @xmath171 & @xmath172 + ngc 1222 & @xmath173 & @xmath174 & @xmath175 & @xmath176 & @xmath177 + sbs 0335 - 052 & @xmath178 & @xmath179 & & @xmath180 & + ic 342 & @xmath181 & @xmath182 & @xmath183 & @xmath184 & @xmath185 + vii zw 19 & @xmath186 & @xmath187 & @xmath188 & @xmath189 & @xmath190 + ngc
1741 & @xmath191 & @xmath192 & @xmath193 & @xmath194 & @xmath195 + ii zw 40 & @xmath196 & @xmath197 & @xmath198 & @xmath199 & @xmath200 + mrk 8 & @xmath201 & @xmath202 & @xmath183 & @xmath203 & @xmath204 + mrk 86 & @xmath205 & @xmath206 & @xmath207 & @xmath208 & @xmath209 + ngc 2903 & @xmath210 & @xmath211 & @xmath212 & @xmath213 & @xmath214 + mrk 1236 & @xmath215 & @xmath216 & @xmath217 & @xmath218 & @xmath219 + ngc 3077 & @xmath220 & @xmath221 & @xmath222 & @xmath223 & @xmath224 + ngc 3125 & @xmath225 & @xmath226 & @xmath227 & @xmath228 & @xmath229 + arp 233 & @xmath230 & @xmath231 & @xmath232 & @xmath233 & @xmath234 + arp 217 & @xmath235 & @xmath236 & @xmath237 & @xmath238 & @xmath239 + haro 3 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath244 + pox 4 & @xmath245 & @xmath246 & & @xmath247 & + , white ) and circle with radius ( @xmath250 ) equal to the separation between the `` on '' and `` off '' beams ( blue ) overlaid on an optical ( sdss g ) image of m 101 .
m 101 and m 51 are both larger than the beam separation , which likely results in an oversubtraction when the flux in the `` off '' beam is subtracted from the flux in the `` on '' beam.,width=566 ] ) component , the blue dotted line is the thermal ( @xmath251 ) component .
when a galaxy s sed could not be fit with the inclusion of a positive thermal component , we only fit the nonthermal component ( the thermal flux at 33 ghz is calculated as an upper limit in such cases ) .
galaxies that are resolved at 33 ghz are marked with an * . ]
fluxes according to equations 1 and 17 from @xcite .
the solid black line represents equal sfrs at radio and infrared wavelengths .
most galaxies have higher sfrs when calculated using radio continuum fluxes , which trace more recent star formation than infrared fluxes . ] and 100 @xmath56 data .
all of the galaxies unresolved with the gbt s 23@xmath19 beam are plotted except for sbs 0335 - 052 , which was not detected by iras , and pox 4 , which was not detected at 100 @xmath252 .
the lines that best fit each data set are also plotted .
the galaxies are coded by thermal fraction at 33 ghz .
while the correlation is tighter at 1.4 ghz than it is at 33 ghz , it is still easily seen at 33 ghz .
since the galaxies with the highest thermal fractions all lie above the best - fit line at 33 ghz ( but do nt at 1.4 ghz ) , it is possible that some of the scatter in the correlation at 33 ghz is due to the increased proportion of thermal emission at higher frequencies . ]
vs @xmath73 using 33 ghz fluxes for unresolved galaxies .
@xmath253 is a measure of the ratio between radio flux and total far - infrared flux at a given radio frequency . as in figure
[ fig : rad - ir ] , sbs 0335 - 052 and pox 4 are not plotted .
the red diamonds represent the highest thermal fraction ( greater than 75@xmath13 ) .
the green squares represent galaxies with thermal fractions between 50@xmath13 and 75@xmath13 .
the blue circles represent galaxies with thermal fractions less than 50@xmath13 .
the purple inverted triangles represent galaxies where we were only able to determine upper limits for their thermal fractions .
the light blue triangles represent galaxies that were only marginally detected at 33 ghz , so no thermal fraction was calculated . at 33 ghz ,
the ratio of radio flux to total fir flux is highest when @xmath73 is flat and thermal fractions are high .
these three properties are all indicative of recent star formation .
thus , it is possible that these properties together act as a rough measure of the timescale of the current episode of star formation . ] | we observed radio continuum emission in 27 local ( d @xmath0 70 mpc ) star - forming galaxies with the robert c. byrd green bank telescope between 26 ghz and 40 ghz using the caltech continuum backend .
we obtained detections for 22 of these galaxies at all four sub - bands and four more marginal detections by taking the average flux across the entire bandwidth .
this is the first detection ( full or marginal ) at these frequencies for 22 of these galaxies .
we fit spectral energy distributions ( seds ) for all of the four - sub - band detections . for 14 of the galaxies ,
seds were best fit by a combination of thermal free - free and nonthermal synchrotron components .
eight galaxies with four - sub - band detections had steep spectra that were only fit by a single nonthermal component . using these fits , we calculated supernova rates , total number of equivalent o stars , and star formation rates within each @xmath1 beam . for unresolved galaxies ,
these physical properties characterize the galaxies recent star formation on a global scale .
we confirm that the radio - far - infrared correlation holds for the unresolved galaxies total 33 ghz flux regardless of their thermal fractions , though the scatter on this correlation is larger than that at 1.4 ghz .
in addition , we found that for the unresolved galaxies , there is an inverse relationship between the ratio of 33 ghz flux to total far - infrared flux and the steepness of the galaxy s spectral index between 1.4 ghz and 33 ghz .
this relationship could be an indicator of the timescale of the observed episode of star formation . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.